I hope I'm not being too controversial when I say I've stopped even looking at the AI models.

I don't think they add any value, in fact I've noticed they exaggerate some of the models in built biases. For example, we are plauged with models generating northern blocking much more often than it actually happens; the AI models are also always the first to do that. I wouldn't be surprised if they are also very good at underdoing cold T850s in winter and hot T850s in summer too. The ensembles are a real world, physical pattern finder that seems to be alot more useful to look at than the AI stuff.

Again might be wrong, its just intuition after using these things for a year or so. 

There's an interesting article out from NOAA re. the new AIGFS which includes info about their reliability.

https://www.noaa.gov/news-release/noaa-deploys-new-generation-of-ai-driven-global-weather-models  

And this is fascinating from the ECMWF camp. 

https://charts.ecmwf.int/products/plwww_3m_fc_aifs_wp_mean?area=Northern%20Extra-tropics¶meter=Geopotential%20500hPa&score=Root%20mean%20square%20error 

I rate AIFS when it comes to its 500hpa charts not so much it's 850s. It had a good summer often picking up the heatwaves first and more accurately than others.  It isn't perfect though far from it.

The problem is the evaluation criteria though, the AI doesn't solve physical equations; it just tries to achieve the best possible validation statistics according to whatever they happen to be. One can easily imagine it fulfilling those criteria better than the normal models over the whole domain but at the cost of producing nonsense at a local level that still averages out to something that looks good. 

Yes P14 is straight out of the January 1987 cold store, lovely stuff!

Interesting news- do you have a link for that as I can't locate on Meto site?

45 years ago- end of 1980- saw an unusual easterly block at the start of November and a strong block with easterlies in the second half of the following February with snow under both set-ups. December I recall was milder than November!

It is on Youtube

Feb 86 was a strong block...coldest Feb for years 

Specifically here:

https://www.youtube.com/live/4GlOt47PzoQ?si=Ez90OvIisXqaWrFa&t=1213

Lots of caveats on the 45 year comment!

"Some of our chiefs here have been having a look at this, and it does look like it's one of the strongest blocks that we've seen of this type, in this area, for this time of year, that has developed over the past 45 years"

A few caveats in that!!

🤣🤣🤣 Well why not, tis the season to be jolly.

Cheers for the laugh 😂 

That must be close to a world record single paragraph caveat count. Very much in keeping with their extended written forecasts 😜

GFS op brings in a second cold wave around +216

Maybe there’s a bonus for most caveats? 😉

I have to say that when I watched that I didn’t see it as caveats, just stating precisely what the record was, bearing in mind they also talked about the likely longevity (at least 2 weeks), which also ties in with what Darren has been saying in relation to the ECM longer range charts.

Oh, if only, if only!

The Canadian GEM shows plenty of snow falling and settling during the day too, but of course it's the GEM - I'd take those charts with a massive dose of salt even if they weren't 6 days away. It's good to look at, anyway.

https://images.meteociel.fr/im/95/12252/gemfr_16_150imo2.png 

There are, as you'd expect, considerable differences between the models and between their runs with regards to that upper cold pool and whether it tracks to the south or tracks over us - the cold pool is the key to seeing snow on Christmas Day, but even 24 hours out we'll not know for sure...

https://www.wetterzentrale.de/maps/GFSOPEU12_282_1.png 

One thing I have noted from the ECM op runs over the past week or so is that once they detected the pattern change in the days leading up to Christmas just over a week ago, they have stuck with it pretty solidly ever since. From memory, it took a little longer for GFS to get properly on board with this and it initially wanted to continue the unsettled spell.

However this change has come about, it sure will make a change from seemingly neverending wet weather. 👍

The presenters in today’s Met Office Deep Dive were categorically adamant the brief technical SSW event late Nov has not influenced what is happening now.