The models are all initialised with a mixture of actual and estimated data. There are estimates because there are more data points in the models than there are observations/data available and that is why the data ensembles are generated, to test how critical changes to the estimated opening position are to the subsequent evolution.

The models are programmed with the best understanding of the physics of how the atmosphere works.  They don't have any particular weather phenomenon 'programmed in': when you introduce a forcing into the atmosphere you get a response.  The models will attempt to reflect the forcing happening in the stratosphere but will be no better than they are at handling anything else.

Agreed. Even if there was a signal for an Easterly influenced colder evolution in the models from mid-January it would be deep FI stuff and therefore would have slim chances of verifying.

Conversely of course, the pressure rise over Europe and low pressure to our NW is also in FI, so there's doubt about that happening also

I think all you can say about the current MO is that things look fairly benign and mild initially with a likely move to colder anticyclonic conditions, with no real signals for a change in that pattern as we head into January.

Well that surprised me with the snow depths. I didn't do a count but the majority seem to show lying snow in my area 11th Jan at 6am. Some have a lot of snow.

Agree with all this.

I indulge sometimes in a bit of what I think others have hitherto called 'hopecasting' and I don't mind admitting that I'll do the same in the future, because I do like a cold, snowy spell in winter. But I'm well aware that the intricacies of atmospheric dynamics will never reveal their entire, or even most of, their secrets to me. 

Another recurring issue here is the excessive focus on the output from GFS.  Because it runs four times per day andmakes available far more information it receives far more attention.  That doesn't make it right or any better than other model output, indeed the performance stats suggest they're generally just behind UKMO and ECM.

I noticed that issue last winter. Obviously I don't know the detail of how the model works but that website provides precipitison charts which show 'rain and snow' as one output and I wonder whether they're treating rain and snow as just snow when it is converted to snow depths?  Certainly I took the snow depth with a very large pinch of salt unless all the other parameters confirmed it was just snow falling.

I was watching a UK Met Office weekly YouTube program earlier last week (excellent weekly show) and they showed two models showing the possible Christmas pattern. The 'American' model (which I assume is the GFS) and another unnamed one, which looked curiously similar the what the ECM model was showing at the time. Interestingly, they took the side of the (presumably) the ECM forecast pattern, which showed HP to the SW and a diverging WNW flow, over the GFS, which had lower pressure to the SW and a more southerly airflow. 

Also interestingly, while none of the two models were exactly spot on (keep in mind this forecast was a week before Christmas) and in spite of the Met Officer's own instincts, the GFS proved to be far closer to the mark in the end. 

Martin

Looks like there is something not quite right with the way they calculate this parameter. Also now I can only go out to 6th Jan...pretty convinced I could see out to the 11th earlier! 

Yes a cooling trend, although the source of this by the end of the ENS seems to be a strong Westerly type. 

For the lay, self included, it's also convenient as we can access the ENS set easily through TWO, giving 22 outcomes to consider, including trends and clusters. I don't consider the higher time resolution to be useful- when doubting beyond T+120, a 6 hour update is of limited interest. I tend to use the 0z mostly. 

Furthermore, GFS comes into view more quickly in the morning- when I start work, the 0z is usually fully rolled out. The ECM may be, but more often than not it's still updated only to 72 or so.

Regarding your last comment, didn't there used to be a regular monthly comparison of the 3 models? ECM was usually the most accurate retrospectively. I can't remember if it was one of Gav's- back three years or so. Any recollection?

I think UKMO is available before GFS has been updated fully but I do tend to start with GFS.  

The verification stats are available online but I don't have the link on my iPad.

Don't forget that the op only goes to day 10, which would be Jan 6th.

Same happened 21st January 2015 when we were going away!  We wouldn’t have got further than the end of our street if a snowplough hadn’t conveniently led the way.  My winter holidays are badly timed!  I miss all the lovely snow and our flights get delayed.  

LOL. Please don't go away in the spring.

ECM 00z ensemble suite for London:

The Op sat amongst the colder options but within the ensemble range. Limited milder options later in the run with the main cluster showing near-normal temperatures but plenty of cold members and a reasonable but small cluster showing very cold weather.

So I thought I’d try February 18th for the next one!  Watch this space! 

Day 5 verification stats by 500hPa anomaly:

http://www.emc.ncep.noaa.gov/gmb/STATS/html/new_acz5.html

EC running top both North and South Hemisphere

But still the nastiest month of the year in my opinion. Rare to see any sign of growth until late April, early May, here at least. 

The edging back in the UKM long ranger looks sensible IMO.

I also think pushing back the cold makes sense at this stage of the game. TBH a lot of the winter forecasts are quickly turning into fire and forget publications but as we're still in December it is far too early to call time at the bar.

thank you 

most helpful 🙂

That is quite some disagreement at relatively short range (although the forecast ties in wi the UKMO and Icon output).

Our older members will remember those days fondly....... 🙄

Whatever happened to SWZ ?