Winter 2011-2012

[tacoman25]

Here is the November composite preceding -ENSO winters that heavily favored cold in the west/warmth in the east. No correlation at all to the following winter patttern.

[OKpowdah]

I think the October NAO correlation has merit and therefore you can have some meaningful correlation with the winter temps. But I do agree "October temps=winter temps" here is a very dubious connection. Especially since the October NAO correlation is actually an inverse correlation...and the temp correlation is quite small to the NAO in October as well.

I didn't say anything against the October NAO correlation. There is definitely merit there, and as I've said before, I like the use of an inter-annual oscillation in the NAO, which would suggest the SON NAO has some correlation with the DJF NAO.

But yeah, just as you said, since wavelengths are shorter in October than the winter, you don't get the same temp correlation to the NAO.

[tacoman25]

I think the October NAO correlation has merit and therefore you can have some meaningful correlation with the winter temps. But I do agree "October temps=winter temps" here is a very dubious connection. Especially since the October NAO correlation is actually an inverse correlation...and the temp correlation is quite small to the NAO in October as well.

I think as last year demonstrated, it's more during extreme years where the correlation exists.

Looking closer at individual years, it seems that -ENSO years that don't feature big western troughing in October are most promising for the following winter in the east. -ENSO years that do feature big western troughing are less promising for the east. That is the most consistent pattern trait. So in that sense, I do think the October longwave patterns can provide some clues to what patterns may dominate the following winter, though these are just likelihoods, not guarantees.

[HM]

First of all, the amount of cold October years vs. warm October years isn't equal. Running cold winters and warm winters backwards to their October temperatures doesn't work, if you are trying to show October temps and the following winter. If you do it the other way, the signal will modify more. The warm October years will trend cooler across the Northeast and the cold October years will trend warmer across the Southeast (this also means that both sets of years bring a cold winter to New England now). This is because, ENSO is the dominate signal and the October temperatures are meaningless.

Here is how the years should be broken down:

Warm Octobers across the Northeast: 1903, 1908, 1932, 1942, 1949, 1955, 1956, 1970, 1971, 1985, 1995, 2007 and 2010

Cold Octobers across the Northeast: 1917, 1933, 1964, 1966, 1988 and 2008

Unfortunately, there aren't enough years; 13 vs. 6 isn't exactly fair. There are inherently going to be more basin-wide La Niña events in the higher data set which will likely affect the winter temperature signal to favor more of a Southeast Ridge.

Warm October Temperatures:

Their Winter Temperatures:

Since 1950:

Cold October Temperatures:

Their Winter Temperatures:

Since 1950:

Notice that when you focus on the cold events since 1950, the Northeast temperatures are identical, along with the northern Plains. The Southeast Ridge is the only difference between the two sets and that is likely due to the lower amount of years and healthy La Niña events in the cold October years.

[tacoman25]

HM, my focus was not so tightly on the NE, but on the general pattern across the CONUS. For instance, you list 2010 as a "warm" October in the NE, but if you look at the overall pattern across the nation, the NE was cooler and the West blowtorched - no sign of typical troughing that often sets in during -ENSO Octobers.

Oct 1903 also featured dominant western ridging, another year that went on to a cold winter in the east.

[HM]

Okay...

But then I can bring up 1995-96 which had the NW trough that you are looking for and a warm Northeast profile and yet it lead to a cold winter here. Same can be said for 85-86. The point is: there are simply not enough years to make any connection.

[NYsnowlover]

Okay...But then I can bring up 1995-96 which had the NW trough that you are looking for and a warm Northeast profile and yet it lead to a cold winter here. Same can be said for 85-86. The point is: there are simply not enough years to make any connection.

Do you think that if it had not been for all of the clippers, the NE would have not seen the snowfall amounts that it did?

[Ginx snewx]

Do you think that if it had not been for all of the clippers, the NE would have not seen the snowfall amounts that it did?

If my Aunt had balls she would be my Uncle. Seriously, clippers are part of the pattern, if they had not occurred yes there would be less snow, kind of obvious though.

[NYsnowlover]

If my Aunt had balls she would be my Uncle. Seriously, clippers are part of the pattern, if they had not occurred yes there would be less snow, kind of obvious though.

LOL, Just one more question, do you think we will see more clippers this year, than last year?

[uncle W]

storm tracks and amounts of precipitation are equally important...These are the Octobers and Novembers before NYC's snowiest winters...

[tacoman25]

Okay...

But then I can bring up 1995-96 which had the NW trough that you are looking for and a warm Northeast profile and yet it lead to a cold winter here. Same can be said for 85-86. The point is: there are simply not enough years to make any connection.

I have to disagree. There is a pretty good sample of years, providing a bigger dataset for analysis than many of the other correlations that are often made on here. There is a correlation and an inverse correlation that agrees. -ENSO Octobers where the West is warmer than the East are more promising for the East than the opposite. The stats do back this up, as I have shown. Even your maps including 1903 and 2010 in the "warm" dataset show that when the October preceding is colder in the East, the East overall tends to be colder than the West the following winter.

[tacoman25]

To illustrate the October correlation better, take a look at this. The first map is all -ENSO Octobers that were clearly colder in the East than the West overall. The next map is the following winters. Then we have all -ENSO Octobers where the West was clearly colder than the East. And the following winters.

[ORH_wxman]

To illustrate the October correlation better, take a look at this. The first map is all -ENSO Octobers that were clearly colder in the East than the West overall. The next map is the following winters. Then we have all -ENSO Octobers where the West was clearly colder than the East. And the following winters.

I think your composites are being affected a lot by pre-1950 years. There's still a difference between the years, but its definitely a weaker one when using post-1950 stuff.

[tacoman25]

I think your composites are being affected a lot by pre-1950 years. There's still a difference between the years, but its definitely a weaker one when using post-1950 stuff.

Sure, it's stronger when you add in the the pre-1950 years (any reason one wouldn't in this case?). But when you also consider the inverse correlation that matches, I think the overall October/winter connection is fairly substantial.

[ORH_wxman]

Sure, it's stronger when you add in the the pre-1950 years (any reason one wouldn't in this case?). But when you also consider the inverse correlation that matches, I think the overall October/winter connection is fairly substantial.

The overall cold signal is stronger in the those years so it would help enhance the "cold winter for east" years. I just do not think the conncection is as substantial as you are trying to portray. I do think there might be something to the "west is warmer than east" idea in October during Ninas...perhaps this is an underlying result of some larger scale patterns that correlate to something in the winter...but I think a much more useful correlation to temps would be the AO/NAO in October rather than using sfc temp splits.

You'd also want to look at the individual years to make sure a few aren't skewing the result. A composite can be useful, but you want to make sure a majority of the years agree.

[tacoman25]

The overall cold signal is stronger in the those years so it would help enhance the "cold winter for east" years. I just do not think the conncection is as substantial as you are trying to portray. I do think there might be something to the "west is warmer than east" idea in October during Ninas...perhaps this is an underlying result of some larger scale patterns that correlate to something in the winter...but I think a much more useful correlation to temps would be the AO/NAO in October rather than using sfc temp splits.

You'd also want to look at the individual years to make sure a few aren't skewing the result. A composite can be useful, but you want to make sure a majority of the years agree.

This would also make the entire country colder (though the difference really isn't substantial). What I was looking for is temperature gradients across the country, the exact anomalies don't matter. I am not trying to portray the connection for anything other than what it is.

Yes, a few years can somewhat skew a composite, but that is always the case. In this particular case, the sample size and anomalies are large enough that I think it's actually less of an issue than most times people make composites.

[Snowstorms]

It seems like the La Nina is starting to take effect somewhat. We see a decent SE Ridge developing across the Atlantic through the next few days and stays consistent atleast through much of next week. Models develop a Colorado Low of some sort late next week followed by a Clipper then a Cool down across the East.

PNA stays negative through next week with a neutral NAO and slightly positive AO.

Thus far the SST configuration looks very similar to 2008 across the Atlantic/Pacific. The cold anomalies across the GOA are very similar to that of this year along with a well defined -PDO. However across the Atlantic the SST anomalies don't resemble 2010 but resemble 2009 somewhat. Other similar years in the last few years include, 2004, 2002 and 1999 and this is only speaking for the Atlantic. Surely enough the PDO stayed positive through 2006 before going negative back in 2007.

Notice most of the analogues I included featured High Tropical Storm activity across the Atlantic.

BTW the MJO is currently forecasted to move towards phases 6/7 which is not favorable for La Nina development. We'll see if any Kelvin waves develop but we should be seeing another burst of Trade Winds across the Nina regions next week with the SOI remaining positive with the GLAAM near neutral currently associated with the unfavorable MJO and with the GWO in phase 4. The Subsurface cooling seem to have expanded Westward towards the Western Nina regions lately but I haven't seen any significant cooling lately but I expect it to resume come Mid October or so.

[ORH_wxman]

This would also make the entire country colder (though the difference really isn't substantial). What I was looking for is temperature gradients across the country, the exact anomalies don't matter. I am not trying to portray the connection for anything other than what it is.

Yes, a few years can somewhat skew a composite, but that is always the case. In this particular case, the sample size and anomalies are large enough that I think it's actually less of an issue than most times people make composites.

Look at the two maps I posted. The first one is completely unspectacular, the 2nd shows the propensity for a more robust SE ridge but effects New England/Northeast very little in the scheme of things.

It might be of importance to further south in the Mid-Atlantic, but overall for which one is more likely to produce below average temps in the northeast/New England, I'm not sure this says anything.

[HM]

I have to disagree. There is a pretty good sample of years, providing a bigger dataset for analysis than many of the other correlations that are often made on here. There is a correlation and an inverse correlation that agrees. -ENSO Octobers where the West is warmer than the East are more promising for the East than the opposite. The stats do back this up, as I have shown. Even your maps including 1903 and 2010 in the "warm" dataset show that when the October preceding is colder in the East, the East overall tends to be colder than the West the following winter.

Yes but when we were cherry-picking, it seemed to have clear outliers and contradictory analogs. That is all I am saying, which would make me, anyway, unsure of their value when trying to forecast the upcoming winter. Let's say I get a profile of cold west and warm east this October...in the back of my mind I am thinking, well heck there was 85-86 and 95-96 that bent the rules, despite the other years.

I still think that 2010 and 1910 look very similar and that they are in their own class. I also think 66 is cold everywhere across the CONUS outside of the Chinook areas. All of these years should be rejected.

It's just seeing 99-00 in the cold years and 95-96 in the warm years --- just doesn't sit right with me.

[tacoman25]

Yes but when we were cherry-picking, it seemed to have clear outliers and contradictory analogs. That is all I am saying, which would make me, anyway, unsure of their value when trying to forecast the upcoming winter. Let's say I get a profile of cold west and warm east this October...in the back of my mind I am thinking, well heck there was 85-86 and 95-96 that bent the rules, despite the other years.

I still think that 2010 and 1910 look very similar and that they are in their own class. I also think 66 is cold everywhere across the CONUS outside of the Chinook areas. All of these years should be rejected.

It's just seeing 99-00 in the cold years and 95-96 in the warm years --- just doesn't sit right with me.

Well, all I can say is that we could nitpick these same sorts of things with any correlation. There are always going to be years that don't follow the correlation. You're not going to have 100% or rarely even 90% correlations with weather/climate.

[weathafella]

I have been working on the winter forecast but the heat and humidity this weekend tells mad em too lazy. Hint: mix of analogs with some spectacular and some awful make the confidence this year lower than normal.

[HM]

Well, all I can say is that we could nitpick these same sorts of things with any correlation. There are always going to be years that don't follow the correlation. You're not going to have 100% or rarely even 90% correlations with weather/climate.

I understand this but I am saying I don't agree with the years that you selected either, which may possibly weaken or strengthen the correlation. The other thing is: I don't trust the data before 1950, despite recent advancements in reconstructing our climate history.

[tacoman25]

I understand this but I am saying I don't agree with the years that you selected either, which may possibly weaken or strengthen the correlation. The other thing is: I don't trust the data before 1950, despite recent advancements in reconstructing our climate history.

Why don't you agree with the years I selected? They all met the criteria: -ENSO and dominant ridging/warmth in West or dominant ridging/warmth in East.

[ORH_wxman]

Why don't you agree with the years I selected? They all met the criteria: -ENSO and dominant ridging/warmth in West or dominant ridging/warmth in East.

ENSO data prior to 1950 is subject to large errors as CPC shows on their site. There was a very significant narrowing of the margin of error through the 1940s.

[snowNH]

Will, when are you planning on issuing your winter outlook?

[ORH_wxman]

Will, when are you planning on issuing your winter outlook?

I never issue one. I don't believe my skill is high enough to do one in good faith. I do throw out snowfall amounts for fun sometime in late October or early November though.

[tacoman25]

ENSO data prior to 1950 is subject to large errors as CPC shows on their site. There was a very significant narrowing of the margin of error through the 1940s.

That's fine, I'm not worried about exact anomalies. All of the years I listed were pretty clearly -ENSO, some weaker and some stronger. The data is not as good, agreed, but in this case that isn't too important.

[HM]

There are some years where I think it is inconclusive or too "hybrid" and should be tossed. This is because we are using surface temperature anomalies and trying to figure out where the troughs/ridges are--which may not always be clear-cut.

I think 1910 and 2010 look very similar yet you have them in opposite categories. Either, both of them should be tossed with their widespread warmth or you should move 1910 to the other selection (I know at 500mb, 2010 had a trough in the East, despite the warm temperatures). For the opposite reason, I think 1966 had a trough in the East and ridge along the West Coast but the temperatures were below normal in the West, outside of the northern Plains, too. It just isn't that clear PNA that you want to see.

In 1924, 1956 and 1973, the warmth was in between the East and West coasts...well at least the heart of the anomaly was. But all of these had ridge anomalies in the East and troughs in the East Pac/west coast/west Atlantic. But these 3 years don't seem to look like an RNA signal.

If you want to show a trough-ridge relationship, fine; but, surface temperatures are not always going to represent that as we saw with 1966 and 2010.

Finally, the lack of a relationship post-1950 is a bit worrisome if you are looking for a meaningful correlation. But, in the end, maybe I'm just nitpicking.

[Typhoon Tip]

There are some years where I think it is inconclusive or too "hybrid" and should be tossed. This is because we are using surface temperature anomalies and trying to figure out where the troughs/ridges are--which may not always be clear-cut.

I think 1910 and 2010 look very similar yet you have them in opposite categories. Either, both of them should be tossed with their widespread warmth or you should move 1910 to the other selection (I know at 500mb, 2010 had a trough in the East, despite the warm temperatures). For the opposite reason, I think 1966 had a trough in the East and ridge along the West Coast but the temperatures were below normal in the West, outside of the northern Plains, too. It just isn't that clear PNA that you want to see.

In 1924, 1956 and 1973, the warmth was in between the East and West coasts...well at least the heart of the anomaly was. But all of these had ridge anomalies in the East and troughs in the East Pac/west coast/west Atlantic. But these 3 years don't seem to look like an RNA signal.

If you want to show a trough-ridge relationship, fine; but, surface temperatures are not always going to represent that as we saw with 1966 and 2010.

Finally, the lack of a relationship post-1950 is a bit worrisome if you are looking for a meaningful correlation. But, in the end, maybe I'm just nitpicking.

Funny I was just mentioning this in the other thread but this reminds me of that. ... Folks, I'd be cognizant of notion that the EPO/AO/NAO multi-decadal signature has gone negative and if historical records server any predictive measure the sign will increase in negative magnitude before rising again spanning the next 2 decades.

That's going to insert a pretty substantial correction to just about everything.

Personally I'd focus on ozone dispersion in the p-v with a bent/nod toward solar activity enhancing residence decay as a wild card, and otherwise figure like last year's -AO domination in DJF - that's favored as a background canvas.

I still like the gradient idea event though ...I know, I know, that is not a prediction. Gee wiz

[Snowstorms]

There are some years where I think it is inconclusive or too "hybrid" and should be tossed. This is because we are using surface temperature anomalies and trying to figure out where the troughs/ridges are--which may not always be clear-cut.

I think 1910 and 2010 look very similar yet you have them in opposite categories. Either, both of them should be tossed with their widespread warmth or you should move 1910 to the other selection (I know at 500mb, 2010 had a trough in the East, despite the warm temperatures). For the opposite reason, I think 1966 had a trough in the East and ridge along the West Coast but the temperatures were below normal in the West, outside of the northern Plains, too. It just isn't that clear PNA that you want to see.

In 1924, 1956 and 1973, the warmth was in between the East and West coasts...well at least the heart of the anomaly was. But all of these had ridge anomalies in the East and troughs in the East Pac/west coast/west Atlantic. But these 3 years don't seem to look like an RNA signal.

If you want to show a trough-ridge relationship, fine; but, surface temperatures are not always going to represent that as we saw with 1966 and 2010.

Finally, the lack of a relationship post-1950 is a bit worrisome if you are looking for a meaningful correlation. But, in the end, maybe I'm just nitpicking.

1955-56 and 1974-75 seem like reasonable analogues ENSO wise and PDO, etc wise as well. Given the lack of data prior to 1950 I think analogues post 1950 should not be used wisely or for conclusions in Final Winter forecast calls and such. I'll agree some analogues like 1917-1918 seem reasonable and compare quite respectively to this year but we must not take advantage of that analogue and other analogues.

The closet analogue in my opinion in resemblance to this year is 2008-09. 1977-78 despite being a Nino seems like a reasonable analogue, dont you think?

BTW the MJO is expected to remain in phases 5/6/'7 for the next 1-2 weeks and this does not seem so great for the current La Nina. The Trade winds however remain strong across the Western Nina regions but I'm concerned about any Kelvin waves trying to develop and moving Eastwards. The GWO also remains unfavorable between Phases 4/5 and currently allowing the GLAAM to go slightly positive recently. The PDO has also experienced minor warming in the past 2 weeks or so.

Any ideas/thoughts on the current MJO phase?

IMO I think the cooling should resume more substantially come Mid-Late October with perhaps a late peak.