2025-2026 Winter Speculation Thread

[Carvers Gap]

23 hours ago, John1122 said:

Obviously taken with Mount Everest made of salt, but I saw someone post the euro seasonal and it was basically normal to AN snowfall across the region. More AN than normal, at least based on 1991-2020 30 year normals. 6-12 inches along and south of 40, 12-20 inches north of 40 into SE KY and SW Virginia. 

Yep.  Its 500 and surface maps don't correlate to the snow maps.  Temps and 500 maps correlate better than last month.  Generally, a very warm pattern for most of winter.  500 is more opportune for November and December if I remember correctly.  The entire winter run was a bit touchy.  I was really surprised to see that snow map circulating as I am not sure that I have access to it or just haven't looked!  FTR, I don't think it is right at 500.  The CFSv2 seasonal has been spitting out some fun 500 runs at times.

[John1122]

I really can't recall the last time a seasonal map didn't show AN temperatures at long leads. Even when we ice boxed last January and were solidly BN for Jan 2024, I'd be surprised if modeling around this time of year didn't show AN temps for January both years. 

[Daniel Boone]

52 minutes ago, John1122 said:

I really can't recall the last time a seasonal map didn't show AN temperatures at long leads. Even when we ice boxed last January and were solidly BN for Jan 2024, I'd be surprised if modeling around this time of year didn't show AN temps for January both years. 

We need to find those Archives. I'm pretty sure they did. I think they showed a Canonical Nina 500 look as well as Temp. Profile.

[Carvers Gap]

1 hour ago, John1122 said:

I really can't recall the last time a seasonal map didn't show AN temperatures at long leads. Even when we ice boxed last January and were solidly BN for Jan 2024, I'd be surprised if modeling around this time of year didn't show AN temps for January both years. 

Oh yeah, seasonal maps are almost always warm at the surface.  I generally just look at 500 maps and source regions - even for weeklies.  The Euro seasonal and weeklies(mean) almost never see cold at range.  Now, the one thing that does work is to look at the control surface and/or 850 temps on the Euro Weeklies - that gives a look under the hood so to speak.  

The seasonal CFSv2 looked good today.  I almost posted it, but it will likely change tomorrow.  The CANSIPS has shown some skill at range w/ its 500 and temps maps...but only the November run has been accurate, though October can provide some hints.  

[John1122]

I honestly think a lot of it is just built into modeling based on ENSO. I was reading the discussion of the 2024-25 winter forecast from NOAA and it's essentially all based on La Nina that they had forecast to begin in winter, and defaulted to what they would say was typical La Nina 500mb patterns which was AN across most of the East. But they admit during the discussion that weaker ENSO events have far less predictability, and that the forecast models incorporate the much more stable strong ENSO results at times and interpret them in a broadbrush fashion. As always, they said the main drivers last winter, PNA/AO/NAO/MJO were simply not able to be forecast more than 3 weeks in advance.

[Daniel Boone]

15 hours ago, John1122 said:

I honestly think a lot of it is just built into modeling based on ENSO. I was reading the discussion of the 2024-25 winter forecast from NOAA and it's essentially all based on La Nina that they had forecast to begin in winter, and defaulted to what they would say was typical La Nina 500mb patterns which was AN across most of the East. But they admit during the discussion that weaker ENSO events have far less predictability, and that the forecast models incorporate the much more stable strong ENSO results at times and interpret them in a broadbrush fashion. As always, they said the main drivers last winter, PNA/AO/NAO/MJO were simply not able to be forecast more than 3 weeks in advance.

Exactly John. I don't know what the weighted percentage is but do know it's by far highest with the Seasonal Models. Basically, so much so the Forecast will show basically Canonical La nina .

[Carvers Gap]

20 hours ago, Daniel Boone said:

Exactly John. I don't know what the weighted percentage is but do know it's by far highest with the Seasonal Models. Basically, so much so the Forecast will show basically Canonical La nina .

We have a long way to go with seasonal forecast models.  I honestly think human forecasters can hand develop seasonal forecasts which have equal or better accuracy, albeit with some AI help.   For now, there is just too much for modeling to juggle.  There is some rigor with seasonal modeling which is released at the beginning of each season, i.e. winter forecasts which are released at the end of November are an example of this.  Right now, ENSO is incredibly hard to predict at range.  Since so many models, as you and John note, are based heavily on ENSO....they swing w/ those ENSO forecasts.  Honestly, it is probably the right way to build a model, but again...way too many other things to juggle.  As Mr Bob notes, each year is its own analog.   Until we have mathematics which can predict chaos at range, there are going to be limits to the power of AI modeling.  A computer can easily beat a human in chess.  But...climate/weather at the 9-12 month lead is like a game with nearly unlimited moves and options.  The human mind is a better predictive tool when it comes to seasonal modeling at the moment IMHO.

[jaxjagman]

For a Nina pattern the Gulf of Mexico has been dormant,other than TS Barry hitting the Bay of Campeche in late June there has been notta,that hasnt happened since 1991,You'd think we'd have an active S/Jet this winter,severe/cold,who knows

[John1122]

Saw another Met who put out a video about winter, he compared it to 2013-2014 as one his top analogs. That's two different people I've seen make the comparison. He noted current sea ice and snow cover were above the last couple of years and similar to 2013-14 and that the Pacific was similar.