IDO

A variation on the theme where we squeeze another 36 hours out of the cold spell with the low undercutting further south. So we are seeing the options, but D9 on 12z and 18z: Similar Pacific profiles pushing the tPV towards our side of the NH and a westerly flow pushing in. Looking at the spaghetti of the tracking of the lows from the Azores, this is one option and no doubt we will see the GFS op showing something further north. There is little legroom to get it further south than this run.

For London, forecast 2m temps as per the ensembles from GEFS (red is mean): After that, we can see the solid warm-up, which was evident for a couple of days on the GEFS and is now being seen cross-model. 16 days 2m.

So with EC moving in line with the ens, we can see longevity being to D8-9. In that case, I would rather have the low from the Azores further north for a snow event before we head back to the westerly regime. D9: The low undercutting south may lead to some surprises but a big snow event is my pick.

The EC D8 is more like the D9 mean than the 0z op, as expected:

Cold would be nice, but this is the main reason I do not want the GFS to verify; D8-16:

On this run the cold at D9: It is edged east by a mini-ridge, so we even lose out on a snow-to-rain breakdown!

We may see 1-2 days more of those -5c uppers based on this run versus 0z. But the issue remains that inter and cross-model support for the NH profile to reflect something like the following is very consistent: With the Pacific high/ridge kicking out the tPV; it has to go somewhere.

It is much better run on the ESB high/ridge, which will impact the following low that flattened the pattern: 06z: 0z:

The second wave of cold air is delayed on this run versus 0z and the low from the Azores is closer, so maybe the snowline further north on this run: 0z: 06z:

I wouldn't say statistical analysis is as helpful in weather models due to the chaos element, so I do get what you mean, and all that I am really saying is to be wary!

It doesn't work like that: σ = √(∑(x−¯x) ( x − x ¯ ) 2 /n) where n = observation total and x are the data values. They probably do not show the ens runs outside the standard deviation as they are outliers but show the op/control as a point of reference to determine if it is or is not a statistic anomaly.

Technically, an outlier is when a run is greater or less than three times the standard deviation of the mean. I assume these are what Meteociel show.

Most outliers become outliers because their standard deviation increases over time, so a deviation from the cluster earlier in the run eventually leads to the OP becoming a statistical outlier later. Ergo: The EC op likely wronged D7-8, compounded by time to be an outlier at D10. The 500hPa temps showed an outlier at D9, confirming that presumption:

For London, the ensembles confirm the op was an outlier for 850 temps at D10 (shocker!): Manchester: The trend was a rise versus the op, a fall at D10. Also, there is an outlier at 500hPa at D9. But overall not a bad mean, even if the OP is maybe a tad colder than likely on this run. The negative is the synoptic mean at D10: No mean undercut going on there, and in fact very similar to GEFS:

The good news is there are no good models at the moment. All of them are struggling as chaos reigns. My main takes are: 1/The GH is a 4-5 day wonder, and all these shortwaves and stalled northerlies mean that any benefit is much less than that envelope. 2/ The optimistic charts where we had the Arctic high and Pacific high/ridge aiding and abetting a longer cold spell are disappearing, leaving something like this: A Pacific high flattening upstream and dismantling the Asian/Scandi trough. 3/ Longevity assuming status quo with 1+2 will have to come from an NW to SE diving Atlantic sector backed up with wedges and a good dose of luck that these features in the trough act together for cold to the UK. 4/ As seen on the EC, the possible snow event is hazardous. If the low pushes too far north, then we risk the pattern becoming messy and end up with a mess! More runs are needed, and as the Met update suggested there is very little clarity from D7+.

At D8 by the time the front/low has been pushed back down south, we can see the hot zone for snow on this run:

T180, and it's north of the M4 so far.

The sweet spot at D7: Cold air winning by then...

The NH profiles D7 less helpful on this run by D6-7:

On this run, the Azores low is further north, and its initial front looks like it will be rain for much of the south:

My best case scenario was that we could sustain the cold after the GH with a NW to SE jet and wedges. This run is spot-on: The GH has a short half-life, so before the Pacific high build on the Asian-Pacific side, establishing the trough and digging deep was the target.

A much better polar profile helps:

This is the point where the HP cell near Greenland shears off, taking with it the sustainability of the GH and sticking it to us, delaying the northerly: T114: T126: Those yellows disappear in 12 hours as the energy flow splits three ways. It is slower than the 0z. Can the second northerly dig south enough before the lows arrive?

I think it is difficult to be positive about the 0z runs. For example the EC for 19th, yesterday 0z -v- today's: We see the EC moving towards the GFS rather than vice versa. It is no surprise, as GFS had already bought GEM and UKMO before the 0z runs with that surface high interloper. The only good thing is chaos exists post D6-7. So there may be something interesting, though the NH profile means we need to get lucky for that (IMBY).

ECM mean is showing the general consensus: Not bad at all. But the devil will be in the detail. As usual, EC is an outlier at D10; do I even have to say that now?