Poor ecm means has to be said high is to close then sinks not unlike the Op .

http://old.wetterzentrale.de/pics/Reem1921.html

http://old.wetterzentrale.de/pics/Reem2401.html

Of course you realise that in isolation those charts are next to useless? They will always come up with an average of all the scenarios, meaning 50% southerly and 50% northerly can average out at the same things as 100% westerly. THis equals "useless" when viewed in isolation.

A glance through the data shows a very high level of uncertainty re. SLP in the Greenland-Iceland area by day 9/10.

Sometimes they can be useful to see where the Op sits. But this time looks like one of those times where its less useful as there does seem to be a split . The Op was one of the mildest runs for De Bilt. 

Means  absolutely nothing , wouldn't get that concerned with a means chart at day 8 and day 10 . Lets look at 144 no further forward.

from Friday onwards it is for the most part one of the milder solutions. By day 10 there are quite a number of very cold runs.

http://www.weerstationgrou.nl/pluim/

Correct Doc, I find them a waste of time

Yes big split by day 10 but very good news to see that run pretty much the mildest run by that point. Some very cold runs by then ironically I think thats the coldest I've seen for De Bilt at day 10 so far this winter. De Bilt is obviously further east than the UK and we maybe to far west to really benefit from whats coming.

Depends on how you read them. I find them far more useful than the op in giving a sense of trend. Of course, the mean always falls close to climatology the further out you go, but there are always subtle differences, and it is up to the viewer to pick these out, I.F, I note, does this in his evaluation of longer range ensemble outputs.

I agree. When I used to post here regularly I used this chart a lot as it tends to change at a slower rate run to run and as you rightly say it is a useful chart for trends etc. Surely an ensemble mean is the same whether it be in chart or cluster form. I have never really found this chart that far from showing what the end result turns out to be come it's day. FWIW I think the dramatic shift it's taken yonight from this mornings run is out of character and could mean a new signal has been picked up. I would urge caution on anyone anticipating a deep cold spell up to two weeks away. What is more certain is the guaranteed colder period for the New Year weekend and let's wait a while and see where we go from there.

As for them never being far from correct, this was the ten day mean chart for 00z this morning

http://meteociel.fr/modeles/ecmwf/archives/2016121600/EDM1-240.GIF?00 

Or the 9 day mean from 9 days ago?

http://meteociel.fr/modeles/ecmwf/archives/2016121700/EDM1-216.GIF?00 

I could go on but the point is a mean is never as useful as looking at all the data (including the mean). At a glance it may sometimes give you an idea but not always.

Yea it's a difficult one to debate. When I look at it I realise that the actual Synoptics may not look nothing like the mean shows come the day but the general trend of lower pressure to the North and high to the South suggests to me there are many members that suggest that type of pattern over others that might show High to the North and Low to the south. So using that method the overall balance has to point towards higher chances of that type of scenario. However I do not diss the value of the clusters as a major forecasting tool as they I agree can be a major pointer to what may be on the way. I concur that the 10 day mean chart is not a tool that can be used in isolation so after all that I suppose you could say we largely agree.

Yes, you're right Martin. I guess, as long as it's all "proper data", the more you look at the better picture you get (or the more confused!)

I recall back in Feb '13 when a very prominent signal was picked up on the EC mean chart at day 10:

As you say, the op wavers significantly from run to run, while the 'mean' changes at a slower rate, so therefore they prove to be more useful in in helping to pick out overall trends.  Of course, the mean run can change markedly at times if prominent signals are picked up on and these should be always be noted.

There is absolutely no way you can estimate with any accuracy or usefulness the distribution of 51 numbers from just one number (the operational) and the arithmetic mean of that number plus the other fifty.

The mean chart does often prove correct, because most weather is indeed average. But I don't see how anyone can draw any kind of conclusions from it and the operational alone when the weather deviates with any significance from the average.

As far as I can see, Fergusson refers to the mean only in conjunction with his knowledge of the entire 51-member suite. Then it can indeed be a handy way of summarizing output and drawing conclusions. But not for the rest of us, restricted to just one member (the op).

But that is the point. Why would think the certain charts from 51 members are more likely to verify?

If you have been keeping an eye on the mean charts over the last couple of months, you would have seen that they have showed anything but 'average' conditions between day 8 and 10 over this period. Even tonight's 850hPa temp chart shows very unaverage conditions for the time of year, with mid-latitude ridging still quite prominent in the near-Atlantic and a generally cool flow over Ireland & the UK:

 How this transpires out in reality is another story, but a chart like this does show an overall trend.. at least until the next run.

From IF

We shall see how it unfolds. ECMWF Monthly also raises +ve MSLP/GPH anomaly to NW (S of Greenland) into week 2; interestingly it supports GloSea5 on broad idea of a colder than average spell much/most of Jan, with varying NW/N/NE flow (perhaps E'ly later in month); even by week 4 with unusually prominent -ve temp anomalies for southern UK (given the lead time). One way or another, model output looks an awful lot more interesting versus this time a week ago. 

It's a nice update but we've been here before with the ecm monthly it got December completely wrong.

Of course, though I do note his assessment of the latest monthly (posted above) ties in very much with the varying anomalies that tonight's EC monthly shows - which I also note, shows something substantially different than to what they showed in the last update beyond the 2 week period.  

Looks like the 18z has gone Pete Tong by 216h for snow fans, cold though.

http://www.wetterzentrale.de/maps/ECMOPEU12_120_1.png 

How long lasting the cooler weather lasts will depend on how much WAA can be pumped to GL in this timeframe. The Newfoundland low looks quite weak to help this, I'd be looking for heights dropping their to fuel the WAA in the next couple of runs to help sustain the block with a faster pump.

http://modeles.meteociel.fr/modeles/gfs/runs/2016122618/gfs-0-240.png?18 

True and also it looks cold and dry not cold and wintry - so the best we would get is frost and fog.

18Z looks OK to me. Plenty of dry and frosty weather. Seasonal, if not snowy, and IMO better than mild south westerlies in any January.