https://ukwct.org.uk/weather/avnpanel0.gif 

https://ukwct.org.uk/weather/avnpanel1.gif 

https://ukwct.org.uk/weather/ecmpanel0.gif 

https://ukwct.org.uk/weather/ecmpanel1.gif 

It's out to 168 on WZ, but not much happens in those extra 24 hours - a proper battle going on!

That chart has proper cold-easterly potential written all over it: strong Scandi high (1035mb+) with a deep Atlantic low to the west = classic setup for an easterly feed into the UK. If it verifies, London and the East could be in line for streamers off the North Sea and/or snow as Atlantic fronts stall against the cold air. Still a long way out, but it’s definitely one to watch.

www.wetterzentrale.de/maps/GFSOPEU00_165_1.png

Not sure AIGFS is that great? Looks pretty marginal for the south with a lot of warm air getting dragged into the mix from the south, and the -5C isotherm just about hugging the coast. Certainly a long way from the deep cold easterlies we have seen in the past. 

It's lost on the previous page, but it bears repeating - the "AI" pattern-matchers show that given the current setup in previous decades, we'd have been looking at a significant cold and snowy spell. Unfortunately for us it's not the past any more, and with more heat and energy in the system it seems to be harder than ever to get a bona-fide long fetch easterly.

It continues to be fascinating, but while I look upon the "AI" models with interest, I don't honestly expect them to be right with such a prolonged easterly spell. 

Yes there is definitely logic to that assessment.  But we live in hope. Think it's quite a big few days for the AI models, have they got this right over the legacy models. And is it still possible to get a proper old school midwinter easterly?

That's harsh! 2m temps would suggest snow all the way from about 216h even in the South. 

GFS Op - the above LP moves SE-wards and HP builds over Scandi with a SE-ly appearing briefly Wed 21st but the Atlantic re-asserts dominance and by Sun 25th main LP 970mb Iceland plus two secondaries running up the west coast offer severe S-ly gales for Britain. This system again runs SE-wards to the Med but this time HP from the continent links to a ridge from the Azores and by Thu 29th Scotland has HP 1045mb and S England has a cold E-ly. Again this is brief as the HP is mobile and moves S to Wales Sat 31st cutting off the E-ly while N Scotland reverts to NW-lies. Two batches of chill, but nothing sustained.

ECM - like GFS to Thu 29th but the HP fails to develop over N Britain , staying mid-Atlantic and Britain resumes as the area of slack pressure between Atlantic LP and continental HP. Only one spell of a couple of chilly days.

AIFS  - London from Maxima 8C now to icedays continuous from Mon 26th, perhaps a little snow by the 29th. Edinburgh, from 5C now to just above freezing on the 26th onwards, a little rain ca  Thu 22nd  and 29th . But see previous posts as to why AIFS may not be the most reliable model at present.

GEFS- mean slowly dropping but near norm to Mon 26th then definitely cold but that cooling as a result of a big spread of mainly colder ens members incl op and control, though most ens below 0C. Rain but not snow in some ens members at this time in the S, paradox? not really, the colder ens members are dry and the less cold ones wet. Scotland more likely to have snow at that time (snow row >50%) , also a little rain this week

ICON keeps it north west of Scandanavia, while GFS sinks it to southern Europe (albeit less far south than the 18Z). Its crucial we know where this WAA ends up because it determines the shape of the anticyclone and in turn where the cold air is.

Also it’s an easterly, usual pre defined parameters change for cold and snow with continental influence. Get the pattern right and it’s usually guaranteed snow.

The shortwave over NW Russia on the GFS 6z, (again not a factor in the AI models this morning) is still there, just like yesterday's GFS runs, but this time, it develops further and it's track, shunts the block West rather than collapsing it South. ECM also goes with the shortwave (again) which influences the outcome. 

Why would the AI models not pick up on this shortwave? Insufficient data input?

I've been analysing today's 0Z GFS and ECM ensembles to see what they show for the 850mb temperature for SE England for 31st January. The GFS ensemble has 32 members if you include the Op and Control. The ECM ensemble has 51 including the Op (discounting Control, as it seems to be broken, the 850mb temp. being shown as identical to the Op throughout the run).

GFS: Op -9.1   Avg. -5.2   >0  2   -5 to 0  16   -10 to -5  10  -15 to -10  4

ECM: Op -6.0  Avg. -3.6   >0  8   -5 to 0  27   -10 to -5  13  -15 to -10  3 

So the ECM ensemble for 31st Jan is a good deal warmer than the GFS. But for both ensembles, around 50% of the members are in the -5 to 0 range: 16 out of 32 for the GFS and 27 out of 51 for the ECM. The GFS has 14 members (almost 50%) being colder than -5, but the ECM has only 16, so only about one-third.

I wonder if any of the AI model's machine learning has been supplemented with 25 years of forum disappointment data to take into account 🙂

An interesting BBC article on AI weather models, focusing on AIFS, ECM's version:

https://www.bbc.co.uk/weather/articles/cwy6ykp7049o 

An excerpt on reliability:

Just like traditional models, AI models are less accurate the further ahead in the future they're trying to predict - a consequence of the chaotic nature of the atmosphere. Looking 10 days ahead, none of the AI models (or traditional models) were able to offer forecasts considered to be of much use in terms of accuracy.

06z gfs looking tantalising ….

Also by then the climatic long term mean has risen from -2C to 0C on the 850 ensembles but this is somewhat counter acted by the sea temperatures continuing to fall a bit.

The near to mid term trend seems to be for the Atlantic to have stopped any initial cold push too far east of us, so we're left looking for a second/third etc. attempt in deep FI. With the majority of the output showing something other than a clean cold feed unless the jet dies down, amplifies more than expected etc. I think it's fair to go with a chilly looking output longer term with cold air close enough at times for wintriness but no more than that. 

GEMS op from the 0z shows that nicely I think iro the mid term. 

So a vague forecast of cooling down as we head in to the final week of the month, with day time highs IMBY around the 4-5c mark, still some unsettled weather and the risk of wintriness for higher parts further north. Low chance of something colder but as it stands mild and unsettled looks even less likely (unless the tendency to collapse the HP to the east accelerates in which case we might see more mild scatter appearing). So kind of a bit of a 'meh' half way house again, a bit like the Xmas period as a few have said but with less of a direct HP influence over the UK. 

I guess glass half full, albeit not for MBY, the 6z op shows a fair few chances for some transient snow falls for some and ends snowy for the West (away from the far SW other than usual Dartmoor blip). We are guess at that point in winter where the Atlantic weather is generally a fair bit cooler than early December, so LPs wandering around with cooler air to tap in to and the pattern not favouring dragging in too much mild air will lead to some wintriness at times. 

Hopefully from a coldie POV either I'm interpreting things badly or the weather gods will relent and stop finding reasons not to get the properly cold air in and I'll be completely wrong😂