What I realized is that there are two main group's with 86-87 and 91-92 extremely warm over the CONUS and the three remaining yrs having your typical evolution with a -EPO and strong NE pac high with a '13-'14 like downstream response. I still need to do further research regarding 86-87/91-92 but from what i've gathered so far the warmth of 91-92 was a result of the extremely high sun spot activity and weak siberian high in October. 86-87 may have been a result of the QBO increase mid winter and negative AMO preventing heights rises near greenland.
Dividing the original set into only the Smax years with increasing solar(left) or decreasing solar(right) have some implications:
The typical reaction from some mets would be to automatically go with a -NAM/SAM state when a nino/-QBO combo presents itself in wintertime. The composite above debunks those expectations and proves that other factors are at work here. Coming into this winter solar activity will be a key factor in addition to the el nino positioning and atlantic SST pattern. Looking at nino yrs with -QBO shear, Smax and decreasing solar heading into winter you still see the NAO predominantly positive in the DJF period. In my case study -NAO months occurred only 22% of the time in the decreasing years, with increasing solar years averaging a -NAO more often (66% of the time) when their is no isolation for ENSO positioning (i.e modiki v.s classic).
What this tells me is that solar trends this fall are important but must be coupled w/ enso positioning in order to make a good forecast. In addition to this, north atlantic SST's and siberian pressure patterns will be key in determining the winter AO/NAO. These indices will not be so clear cut and NAO volatility is likely this winter because of all these factors(I don't think the NAO will be predominantly negative but closer to neutral/weak positive in the means). The benefit of having the NE pac warm pool is that even if the NAO waits until late to go negative which can happen in nino's, you can still have a very interesting pattern, especially for the interior.
Another thing that i've found is that all of my analog years except 51-52 actually had a WB nino and I don't think that it's just a coincidence that all these yrs had a strong GOA warm pool in the fall. Since this is strongly the case again this year we may be headed down that modiki/more blocky road as well, especially if we see a more poleward then normal u wind line as we get into the fall. But again, even if it becomes a modiki, its no clear cut relationship w/ solar still a mystery.
In summary my prelim thoughts are as follows:1) I am favouring a slightly positive NAO this winter
2) More research needs to be done regarding why the 91-92 and 86-87 years differed so much from the other years in the analog set, especially in terms of NE Pac high strength. If we share too many similarities with these yrs in the fall, a very warm winter could be possible(less likely scenario at this point).
3) Weak ridging along the SE coast is likely this winter, with coldest departures displaced NW of the east coast into the plains and GL's.
4) Weak-Moderate nino is favoured given the downwelling phase of the OKW expected in the near future
5) AMO is likely to remain negative but keep an eye on the sneaky +SST anoms in the davis straits
6) PDO is expected to stay positive as we head towards winter
7) Leaning closer to an el nino modiki as opposed to a classic nino setup
8) An interior storm track is more likely overall
9) Warmer then normal conditions expected along the west coast in addition to dryness.