The 12z GFS run can go straight into the bin as far as the New Year period for Southern England is concerned. Rain and rain after rain and wind after rain and wind if that came off. 

Originally Posted by: idj20 

Yes, I think that is it. There will be fewer grid points in the control than the op. It's also worth remembering that the resolution of the data sets which are downloaded and displayed is not necessarily related to that which the model is run at. For example, the GFS data sets are available at 0.25 deg, 0.50 deg and 1 deg. The 2.5 deg ones were discontinued a number of years ago. They all show the "same" output, but if you set a lat - lon for a non-existent grid point, interpolation will be used. Therefore, it's possible to get slightly differnt values for the same coordinate even when using the same run. The same idea will apply to control runs etc.

Originally Posted by: Brian Gaze 

I've never really seen the point of the 'control run',n. 

Originally Posted by: Chunky Pea 

The point is it's the reference, literally the control. When doing an experiment you need a control group, and that's what we get with the ensembles. It's the base state, the default, the "no tweaks here" setting. In theory it'll be a little bit more accurate than the perturbed members, and the point is that if all the other members are wildly different it shows that confidence is especially low, or "Shannon entropy" (to bring back the proper phrase from a few years back) is high.

It's essential, in other words!

Originally Posted by: Retron 

What about in practice though? I've looked at 'control runs'  a handful of times I always found them to be wildly inaccurate. If I watch a TV forecast from Met Eireann or the UK Met / BBC etc, they will either use the ECMWf & UKMO operationals in their maps. 

Could be worth some research on here though. Think I might measure the ECMWF control vs deterministic / mean over the Christmas to see which prove to be more accurate post day 6 or 7. I would be going into this with a bias, but open to having that challenged. 

Originally Posted by: Chunky Pea 

Another point I'd like to make about the 'control', is that, learning what we did yesterday that the higher res (more real world data) the starting point, the more accurate it is likely to be. So how on earth can the 'control' run be of value if starting from a much lower res starting point? 

Originally Posted by: Chunky Pea 

In practice - well, I've found that when the control and op are in alignment then (subjectively) then it's a good sign that something is afoot. For example, when we have one of our many failed cold spells, you'll often see first the op "goes off on one", repeatedly, then the control joins in, then the rest of the pack follow suit. There have been a handful of times over the years where all three (op, control, pack) have flipped as one, but it's rare. Think of the control as the canary in the mine, so to speak.

Note that this only applies to upcoming cold spells, or the end thereof. I've not monitored it so closely in more run of the mill situations, but I would expect it to be similar.

Good luck with your experiment! 👍It'll be hard to quantify, but even a subjective evaluation will be interesting. You'd need a good couple of weeks' worth of data, I would suggest, ideally with either a cold > warm transition or vice-versa.

Originally Posted by: Retron 

Yes, I think that is it. There will be fewer grid points in the control than the op. It's also worth remembering that the resolution of the data sets which are downloaded and displayed is not necessarily related to that which the model is run at. For example, the GFS data sets are available at 0.25 deg, 0.50 deg and 1 deg. The 2.5 deg ones were discontinued a number of years ago. They all show the "same" output, but if you set a lat - lon for a non-existent grid point, interpolation will be used. Therefore, it's possible to get slightly differnt values for the same coordinate even when using the same run. The same idea will apply to control runs etc.

Originally Posted by: Brian Gaze 

I've never really seen the point of the 'control run', and am increasingly of the mind that operational runs are equally as pointless after about 4 or 5 days. I think I seen somewhere that the UKMO is going to scrap the deterministic run in the years to come and just stick with a sort of ensemble aggregate release, in the style of the GEFS etc,  instead. Good idea in my opinion. 

Originally Posted by: Chunky Pea 

ECM bah humbug.  Horrific but perhaps always the likely outcome

Originally Posted by: The Beast from the East 

Worth noting IMO that this morning's UKMO 00z op run went for a similar evolution to the ECM 12z, but the UKMO 12z this evening has gone back to a colder evolution as we head towards the New Year.

Originally Posted by: David M Porter 

Well, for what it’s worth, the 18z GFS has minima of -17°C in the Central Belt next Friday.

https://www.wetterzentrale.de/maps/GFSOPUK18_228_48.png 

Originally Posted by: doctormog 

Toasty in the far SE! That is bitter, interesting to see such divergence across the country 

Originally Posted by: ballamar 

Thanks to Dew for doing the daily updates and wishing everybody here a very Happy Christmas!

Originally Posted by: fairweather 

I've never really seen the point of the 'control run', and am increasingly of the mind that operational runs are equally as pointless after about 4 or 5 days. I think I seen somewhere that the UKMO is going to scrap the deterministic run in the years to come and just stick with a sort of ensemble aggregate release, in the style of the GEFS etc,  instead. Good idea in my opinion. 

Originally Posted by: Chunky Pea 

WX temps - from a position in week 1 with temps across Europe above norm, week 2 becomes distinctly colder from the northeast - N Scandinavia ultra-cold, freezing temps for most of Germany, and outlying patches as far S as Spain. For Britain, freezing in Scotland, Wales, N England, even a blob on Salisbury Plain.  Pptn in week 1 light from Britain and Baltic northwards plus heavy for E Mediterranean; week 2, Britain and Norway dry but general everywhere else and just touching the N & S of Britain too.

GFS Op 00z - current HP gradually flattening out and moving S-wards; by Mon 30th a large LP 980 mb Sweden is projecting weak N-lies across Britain which are intensified by LPs running across England 995mb Tue 31st and across N France 1000mb Fri 3rd moving on to the Baltic. Eventually there are full-on N-ly gales Sun 5th before a tropical maritime HP (like the present one) moves in from the SW 1040mb Wed 8th onwards.

ECM - Similar to GFS but the N-lies Mon 30th are deflected to the east and the subsequent LPs running across Britain do so more quickly with a hint of a brief ridge between (but that on Tue 31st/Wed 1st looks quite vicious as it passes, 980mb Western Isles.)

GEM - HP continues to dominate; the Scandinavian LP keeps well to the east and only on Fri 1st does the HP decline enough to allow an LP 980mb to run past Orkney with only weak N-lies to follow

GEFS - cooling down tom norm by Mon 30th , then a wild scramble of ens outcomes with a range of +/-8C rapidly established and persisting. There is a tendency for the majority of runs to be cold or very cold to Mon 6th (In Scotland and the north of England the cold runs are very definitely predominant) the mean held up by a number of warm outliers; then the majority switch to warm with cold outliers. General agreement on pptn sometimes heavy w/b Sun 29th, dry before and mostly dry after, snow row figures in this week 50%+ for Scotland and N England, minimal on S coast. 

Happy Christmas to all our readers!

Originally Posted by: DEW