SomeLikeItHot

I am not particularly interested in "better" or "worse" in this comment. Personally I would prefer just cold enough for lots of snow to extreme cold and no snow sp I agree its an upgrade. What I am concern with in the post though was the apparently falsely stated accuracy of the GFS ensembles. I do agree though that examining a single point T850 could be misleading, especially given the gradient that occurs on the 10th -11th, but at the same time I anecdotaly observed similar for other locations in the midlands and south. The north has stayed with a similar pattern. Still though, it is saying that the GFS's ensembles modelling of the low was too limited in its range of how strong it would be and now its had to increase that. It varies one brings in mild cold later then gets quite warm. Others just delay it.

If the GFS 6z today actually verified through the run it would have to give the March 1962 CET a challenge.

I have been looking at the various GFS ensemble runs over the last few days and its curious to note that while the METO and ECM have always been hinting at the possibility of later and weaker shot of cold, that the GFS op and with strong ensemble support were refusing to have a bar of it.Certainly in terms of the Op run it appears the GFS has modelled it better. However when you look at what has been going on with its ensembles you have to wonder a bit. As we get closer to the cold on the 11th-13th rather than the spread in the ensembles reducing it has instead grown, suggesting that perhaps the cold plunge is not nailed on as tight as it suggested earlier. So for my area - as that is what I saved charts for we see on the 0z on the 6th we get every single ensemble member suggesting at least -12 uppers on the 11th. By the time we move to the 0z today, we are suddenly seeing that there is a possibility the shot will be less cold than before. With 3 ensemble members going for significantly less cold for the 11th -13th. Come to the 6z and we now have 4 members who don't bring cold 11th to 13th. It should be an expected property of ensemble modelling that the uncertainty decreases with time. I am wishing I had saved the equivalent for the ECM but I didn't certainly though yesterday they still showed a small cluster of slower and warmer members.Now that seems to have vanished. EG The below - some spread but broadly in agreement there through the next few days. The Ecm behavior is what we would have expected from an ensemble set. To me this suggests that the GFS ensembles just aren't particularly good at giving a reasonable coverage of the range of future outcomes. We also saw something similar in the run up to the cold spell at the end of Feb. When the ECM with good (but not universal) ensemble support went for a colder solution, the GFS with really solid ensemble support suggested something mild, only for the whole bundle to switch to a completely different grouping later. So I am interested is there verification stats on how often the ensemble range verifies? It seems to me that this suggests that the GFS ensembles just aren't broad enough. To me its acceptable for an op run to get an evolution wrong say 5 days out, so long as the ensembles support a strong possibility of the alternative, but having an ensemble miss the alternative whatsoever at that range is more concerning.

Wow the 6z is certainly a very cold run, -10 uppers appearing in some parts of the UK at five different times (including this Monday) out into FI.

Given that a couple of days ago you were saying it was only snow for the highest mountains of scotland, a considerable upgrade! As those far more knowledgeable than me have been saying we won't know what snow this sort of convective showers will bring till much closer to the time.

We will see. At the moment, "downgrades" are staying well withing the range of what the ensembles have been saying for the last day or so. If anything the cold for the period 13th to 17th of march has been upgrading if you look at the ensemble mean. If you have believed the op run as the literal truth then maybe there is a scope for seeing this as a downgrade. If you look at the range the ensembles bracket around it then its just an expected variation. Anyway onto the 6z which looks to be an upgrade for the Sth of the UK out to +120

I am struggling to see what down grades people are talking about except either in their minds or in FI? The initial 36 hours of extremely cold uppers (< -10 T80sseems to be more or less the same maybe a couple hours shorter but then longer lasting with -6 to -8 T850s for most of the UK? I think the snow progged by the GFS has changed but that was always likely.

Lol. The tip of hottest march ever is from a person from brighton tourism. Yeah that's a good source!

All this discussion of racing makes me thing that someone should be running a book on when the cold comes - GFS sunday : ECM Monday UKMO (apparently) tuesday. Not at all: Anyone going to offer odds?

Looking at the ECM ensembles, there is still a minority cluster going for cold, but 1-2 days later than the bulk of the ensembles, mean and Op do. So if Mogreps is going for it later (as per the twitter referred to easier). It wouldn't be without some support on the ECM ensembles. I imagine that Mogreps is more biased towards the top group and that is where the UKMO has fallen. However as people have pointed out Mogreps was suggesting SW ly flow well after the ECM op and majority ensembles were predicting a biting easterly during the late Feb cold snap, so it has been wrong before.

compared with the 0z run which proceeds it, 6z is less accurate for the same time forward (ie +120, +144 hrs) but slightly more accurate over 0-120hrs or so than the 0Z to a fixed date. (eg 6am Monday). Yes people are too ready to bin it. My opinion is that its useful in the short term for corrections, and longer term as part of an ensemble view of outcomes. In this case in the short term it slightly corrects towards the ECM (cold slightly later and doesn't sink the high as fast) and this occurs in the timeframe where it is more accurate so its certainly worth looking at.

ha -3 to -4 midday in London on Tuesday according to the 6z. Better put on my thermal underwear.

it is a shame we can't see how the UKMO fits in the MOGREPS suite. The UKMO modifications to it so far suggest that it doesn't sit very well within the set, but that of course could change. The ECM still had quite a cluster (albeit a minority) going for less and/or delayed cold. As I said in an earlier post I find the GEFs strange in that it didn't see any alternatives when the other models could. Be interested to see if the GFS finds any less cold or delayed solutions on this run.

Yes I know but i the 11:55 post it was the Tmax with the ensemble average you posted, which is why Bradythemole said you were using Tmax. In you second post in response to him you posted the average you described (Tmax +Tmin)/2 - which is why they are around 4 degrees lower.

In your previous post you posted the Tmax graph even though I think you were talking about the averages .

Hmm the 6z brings -14 T850s into the north of England at 144 hrs. Can the 12Z make it to -16?

So while we are discussing the incredible GFS mean, does anyone else have the opinion that the GFS ensembles are too narrow in the range of outcomes they project? We saw before the late feb cold spell that 5 or 6 days out they were all against it, while the ECM op and majority ensembles were pointing to it. While the ECM op has flopped around in the last few days, it seems to me that it has always had a cluster of cold members, probably 50% or greater. Indeed it seems poor modelling for the GFS, which we know to be statistically slightly worse than the ECM and UKMO, to not even be able to find a single member greater than -8 T850 for London around the evening of the 11th, while the ECM op is suggesting something more like -2 T850 (just reading off the chart). A lack of cold members in the GFS ensembles implies something like a only 5% chance of something warmer than -8c T850, surely its greater than that.

If the conditions as forecast on the GFS occurred there would be snow well below the highest elevations - if there was any precipitation. AS far as I can see the models aren't projecting a lot of precip with the cold weather, meaning they also show a large diurnal range and relatively high max temps. If the -10 uppers came with precip I am sure we would see snow to much lower levels. All a long way off to be considering that though.

The ECM Op is pretty much at the warmest end of the ensembles post day 5. The Mean is considerably colder. Can't find an archive of this but in the last couple of runs I believe ECM op is at the warmer end of its ensembles - particularly today. Earlier iteration had it cold earlier which quite a number of ensemble members still show. It is perhaps switching between the two possible paths, - Or it could be leading a trend to the milder path. Time will tell.

Sorry I cut the charts for brevity, but it is more than slightly misleading to say that GFS FI is "very Mild in FI " when for 90% of the UK there is -6 to -10 T850s for roughly 200+ hours on the OP charts, a solution supported fairly broadly by the ensembles. Yes it warms up in the last frame but its not representative of the majority of FI.

Over cooked is an interesting term for too cold... I think in terms of weather the GFS is usually over cooked , but interesting to see the similar cold from scandi which didn't hit the uk in the 12z reaching us this time on the 18z and turning into a N to NEly for most of the low res . At this range there is still a of course a very good chance of not verify or being toned down considerably. That said I think at this range the ECM has correctly picked out cold spells including the one last weekend, which it only really "over cooked" in terms of its snow potential. Seems to me considerably increased chances of something cold from early next week from this evenings runs.

If there was notable warmth like occurred in March last year (a cruel trick followed by a month and a half of rain in April/May!) then I would happily look out for it. Mild (low teens) and raining is just not that exciting however. As it stands a decent shot of cold looks about as likely as any real warmth (ie not very). But I don't understand this "its not going to happen" attitude, snow in March is more common than in December.

A Glancing Blow in FI on the 6z

Sunshine? someone has sunshine? Actually I thought I saw it out the window yesterday but it was gone by the time I got outside.

Anything at + 300hrs is not something to build your hopes on, but at least its a nice change for the GFS low res to show something other than a series of lows coming in from the SW.