Model Output Discussion 29th December - Into mid-Winter.

[snowice]

25 minutes ago, January Snowstorm said:

I don't see that it has. The last 3 gfs runs all showed the atlantic gaining ground after the Thursday/night snow window. All has happened as it has dropped the idea of snow coming from the atlantic low on Saturday.

Folks we have learned well before this Winter that if something can go wrong it will when it comes to achieving cold

I see your point however just before the new year Gfs was going for epic cold and the 06z was the chart of the winter only for the 12z to flip the way of UKmo and we know what happen next. I think we all respect the fact that we have to have Gfs on board before we get excited, but because of its wild swings there is more chance it's wrong! 

[IDO]

Looking at the D10 GEFS and about 50% (up from 33%) pushing heights further NEE to get what BA has been hinting at. Similar to this the cluster:

So not a giant step to get to the ECM take. None follow the ECM route, so I suspect this is the middle ground option and the best possible outcome from a GFS perspective, rather than the awful progressive op run?

So still uncertainty post D8, but ECM needs to keep the faith tonight to allow GFS time to catch up, if indeed the ECM surprises us with the exception that proves the rule.

[bluearmy]

We know GFS has a bias to flatten the pattern and we know ECM has a bias to amplify

that can a little or a lot on both models

given where we sit on the eastern edge of a massive ocean with a warm sea current, it's no surprise that GFS is generally right when it goes against the euro models. It's a safe bet.  A broken clock is right twice a day but perhaps think of GFS op as having only a minute hand, given where the uk sits  - it will be in the right place  24 times a day!  It's not as good a model as ECM but it will, more often than not seem to be better for the uk because of its flat bias.  

I won't call the ECM deep cold easterly wrong just yet - waiting for two eps suites to fail to accentuate the ridge days 7/8/9,  but a MLB at our latitude (probably centred just east) remains the most likely outcome. 

[nick sussex]

I think too much attention is being placed on the ensembles and the divergence isn't in FI. We're not talking here of big differences in the NH pattern, both the ECM and GFS agree that the PV will elongate and drop south to the west of the UK.

The ECM at T144hrs is already on its way to developing the easterly because of its cleaner evolution upstream, this is the issue not what lower resolution ensembles have to say at day 8 and onwards.

We are essentially dealing with the different handling of shortwave energy upstream which begins before T96hrs, this then snowballs downstream into either a cleaner Atlantic picture or the messier one shown by the GFS. The UKMO is also not that great because of that shortwave which phases with the energy coming out of the USA at T168hrs. Its still better though than the GFS.

 

[Summer Sun]

West looks best placed for lying snow later in the week especially so northwest Scotland

[knocker]

If you look at the GEFS 8-13 anomaly for 06 you are not a million miles away from the  00z EPS for the same time frame

Edited January 8, 2017 by knocker

[Purga]

2 minutes ago, northwestsnow said:

Has to be said, GEFS 6z are pretty horrid! Eeeeeeeek.

'Horrid' ? I would say that's a bit overstated, the mean is still below average and there are some strong cold clusters as well

The Control run is a stonking easterly as well as the ECM Control run (not shown above) - so 'horrid' wouldn't be my description at the moment.

The ECM shows an easterly. MetO shows trough disription and (by 168) easterlies over half of the UK. GFS is mobile westerly and pretty grim.

So ECM + UKMO v GFS at the moment & still a lot 'to play for' IMO.

 

[The Enforcer]

3 minutes ago, Purga said:

'Horrid' ? I would say that's a bit overstated, the mean is still below average and there are some strong cold clusters as well

 

Crucially, the GFS op is the warmest option for its ubiquitous 'rain day' Saturday. ECM 0z had pleasingly dispensed with the warm sector low and I hope it continues to stick to this approach, regardless of whether there is an easterly afterwards.

[ArHu3]

17 minutes ago, Ed Stone said:

Looking at that spread, I wouldn't give the op, control much credence at all - whatever they happen to show...

That looks like a lot of spread day 8-10 but it's not really, it's just an artifact of the fact that 0 and 360 degrees are the same and the direction of the winds are between nwn and nen

[Jason M]

Re ECM, a big plus point is that it has at least brought the easterly forward to day 9 and personally before I take any interest at all I look for any trend to start moving through the timescales. The downside is that we know it over amplifies at longer ranges and I can't really see any strong support for its solution anywhere else. The ECM with its known tendency to this has the high furthest north, so personally that tells me to treat with extreme skepticism.

The GFS is widely considered to rank just below ECM, but set against this it gives us massively more generous data across a 24 hour period to judge its output. There is a little support within the 6Z suite for the heights to bring in a cold easterly, but by far the biggest cluster is for a return to this winters default pattern. Maybe, we are now reaching the crucial moment in the winter as a whole as were looking to just after mid month. Steve quoted 50 / 50, I'm thinking 80 / 20 against tbh. I also think that if we end up with resurgent heights to our south its going to be a tough road back from there. Kind of like being 2 - 0 down at half time. You can win from that position but you won't usually!

The GFS opp seems way to progressive re a mobile pattern though in the post day 8 period. Euro heights with coolish weather seems the most likely from this point based on experience.

 

 

[northwestsnow]

10 minutes ago, Purga said:

'Horrid' ? I would say that's a bit overstated, the mean is still below average and there are some strong cold clusters as well

The Control run is a stonking easterly as well as the ECM Control run (not shown above) - so 'horrid' wouldn't be my description at the moment.

The ECM shows an easterly. MetO shows trough disription and (by 168) easterlies over half of the UK. GFS is mobile westerly and pretty grim.

So ECM + UKMO v GFS at the moment & still a lot 'to play for' IMO.

 

Maybe horrid is ott lol , fair one

 

Ps is that for london

Edited January 8, 2017 by northwestsnow

[nick sussex]

This made me laugh.

BOTH THE 00Z NAM AND GFS INITIALIZED 5C WARMER THAN OBSERVED AT
850MB OVER WESTERN COLORADO.

So basically the 850 data fed into the GFS was already wrong for that State!

Anyway I'm still continuing my hunt for more info upstream re that shortwave energy.

[Seasonality]

22 minutes ago, SLEETY said:

If it was showing mild would you be taking it seriously then☺they could be trendsetters ,it's hard to know at that timescale ,more runs needed as ever.

Of course. It is amusingly predictable. Cold shows, then a post showing a day 10 or day 15 mean or the ensembles to "prove" the outcome is impossible. Funny it never happens when normal or mild weather is forecast.

The ECM has been playing with colder solutions for a little while now and UKMO seems to be heading down a similar path. Of course it is reasonable to be cautious, cold in the UK is a rare beast after all but the ECM seems more insistent on sniffing out something chilly. Any easterly would be helped by the fact that Europe should get colder again after a brief "warm" up. Ensembles are not always the be all and end all some would have you believe. Especially where unusual outcomes are shown. The op after all has the best resolution and can on occasion be the early trend setter. Caution is the watchword as always but glib dismissal is a form of intentional obtuseness that can be misleading.

[Jason M]

2 minutes ago, Seasonality said:

Of course. It is amusingly predictable. Cold shows, then a post showing a day 10 or day 15 mean or the ensembles to "prove" the outcome is impossible. Funny it never happens when normal or mild weather is forecast.

The ECM has been playing with colder solutions for a little while now and UKMO seems to be heading down a similar path. Of course it is reasonable to be cautious, cold in the UK is a rare beast after all but the ECM seems more insistent on sniffing out something chilly. Any easterly would be helped by the fact that Europe should get colder again after a brief "warm" up. Ensembles are not always the be all and end all some would have you believe. Especially where unusual outcomes are shown. The op after all has the best resolution and can on occasion be the early trend setter. Caution is the watchword as always but glib dismissal is a form of intentional obtuseness that can be misleading.

In fairness it would be difficult to use a mean to disprove a mild set up in recent years as we simply haven't had any meaningful deep cold for over 3 years now!

[bluearmy]

27 minutes ago, Purga said:

'Horrid' ? I would say that's a bit overstated, the mean is still below average and there are some strong cold clusters as well

The Control run is a stonking easterly as well as the ECM Control run (not shown above) - so 'horrid' wouldn't be my description at the moment.

The ECM shows an easterly. MetO shows trough disription and (by 168) easterlies over half of the UK. GFS is mobile westerly and pretty grim.

So ECM + UKMO v GFS at the moment & still a lot 'to play for' IMO.

 

GEFS 06z control isn't an easterly solution? 

[Methuselah]

2 minutes ago, Jason M said:

In fairness it would be difficult to use a mean to disprove a mild set up in recent years as we simply haven't had any meaningful deep cold for over 3 years now!

I don't see how a mean can either prove or disprove anything. Surely it's just another forecast that can only verify or not verify; and that can't be known until after the fact, so to speak...?

[Banbury]

1 hour ago, abbie123 said:

A possible snow event in the south just need The low  further south not a done deal yet with the low to the south west channel low .

That's the 0z run , the LP has now been removed

[ArHu3]

56 minutes ago, IDO said:

Assuming the op and control are similar after D10 then they are massive outliers. Look at the dew point for the region; the mean is positive in FI but the Control is 16c lower the last few days! That is well outside the parameters of standard deviation and IMO outlandish outliers.

The temps not so much, but still coldest solution, but a continental flow or faux cold could keep many of the members colder. Very few going with easterly either:

I cannot take the ECM 0z seriously. 

I never really look past 192h but I haven't seen so much agreement between members before at the day 6-10 range. Anyway don't intent to pay to see ecmwf ensemble charts, so the plume is all I have....

[Jason M]

11 minutes ago, Ed Stone said:

I don't see how a mean can either prove or disprove anything. Surely it's just another forecast that can only verify or not verify; and that can't be known until after the fact, so to speak...?

True, but we can't criticize people for showing mean charts that show mild weather when we had little else for some years as arguably events have proved the poster correct. If come the end of winter we have had several (or even one) strong cold spells then maybe a different conversation (albeit I don't think its right to call people out anyway).

During this winter I've seen lots of potential snow but no real snow. In fact there has been so much potential snow that I've struggled to get out the door some days as the house and car have been buried in the stuff . If GFS is to be believed incidences of potential snow may now lessen going forward . If ECM is believed plenty more shoveling to come....

 

Edited January 8, 2017 by Jason M

[Seasonality]

16 minutes ago, Ed Stone said:

I don't see how a mean can either prove or disprove anything. Surely it's just another forecast that can only verify or not verify; and that can't be known until after the fact, so to speak...?

Elegantly put, as usual Ed.

[Seasonality]

9 minutes ago, Jason M said:

True, but we can't criticize people for showing mean charts that show mild weather when we had little else for some years as arguably events have proved the poster correct. If come the end of winter we have had several (or even one) strong cold spells then maybe a different conversation (albeit I don't think its right to call people out anyway).

During this winter I've seen lots of potential snow but no real snow. In fact there has been so much potential snow that I've struggled to get out the door some days as the house and car have been buried in the stuff . If GFS is to be believed incidences of potential snow may now lessen going forward . If ECM is believed plenty more shoveling to come....

 

When it is the same schtick repeated ad nauseam then some questioning of methodology is more than justified though. Predicting mild in the UK is just betting on the form horse, no skill in it.

Edited January 8, 2017 by Seasonality

[IDO]

15 minutes ago, Seasonality said:

Of course. It is amusingly predictable. Cold shows, then a post showing a day 10 or day 15 mean or the ensembles to "prove" the outcome is impossible. Funny it never happens when normal or mild weather is forecast.

The ECM has been playing with colder solutions for a little while now and UKMO seems to be heading down a similar path. Of course it is reasonable to be cautious, cold in the UK is a rare beast after all but the ECM seems more insistent on sniffing out something chilly. Any easterly would be helped by the fact that Europe should get colder again after a brief "warm" up. Ensembles are not always the be all and end all some would have you believe. Especially where unusual outcomes are shown. The op after all has the best resolution and can on occasion be the early trend setter. Caution is the watchword as always but glib dismissal is a form of intentional obtuseness that can be misleading.

I don't know what you are on about, but a dew point of -15c is pretty unprecedented in recent times for De Bilt. In 2016 the min for Amsterdam was about -8c, ditto 2015:

  Source: here

If you believe that -15c Dew points are close to verifying, then I don't know what i can say! March 2013 was the previous time, -17c. The chart for that day very different:

To get to that extreme then the run must have gone off kilter sometime before, probably before D8. 

Sometimes when an algorithm churns out something as weird as that, you have to ignore it, otherwise all our experience is wasted. If the pattern was different with maybe amplification upstream, then this type of scenario would be slightly more plausible, but it isn't.

Anyway I do not think to point out that the likelihood of a dew point of -15c being implausible as high is "obtuse". 

 

[Methuselah]

16 minutes ago, Jason M said:

True, but we can't criticize people for showing mean charts that show mild weather when we had little else for some years as arguably events have proved the poster correct. If come the end of winter we have had several (or even one) strong cold spells then maybe a different conversation (albeit I don't think its right to call people out anyway).

During this winter I've seen lots of potential snow but no real snow. In fact there has been so much potential snow that I've struggled to get out the door some days as the house and car have been buried in the stuff . If GFS is to be believed incidences of potential snow may now lessen going forward . If ECM is believed plenty more shoveling to come....

 

I'm not criticising folks for posting charts, Jason; this is of course the place for them...It's just that some of the inferences drawn from them can lead to confusion...

Edited January 8, 2017 by Ed Stone

[Jason M]

14 minutes ago, Seasonality said:

When it is the same schtick repeated ad nauseam then some questioning of methodology is more than justified though. 

Hmm, last one from me on this. If events have proved the poster correct how can we question methodology? Maybe people who are saying different should question their own methodology (and that's not directed at anyone in particular). The longer term models have often signposted HLB this season and yet we have had pretty much the opposite of that. The outputs from these might be more interesting, but frankly they have been wrong. In our climate the skill is in picking out the cold spell during the winter months. Lets see if those who always say 'mild' get the outrider correct before we judge. And by that there is little skill in constantly highlighting cold set ups and claiming victory when one finally comes along.

The problem is as always that we need the cold spell before we can judge.

 

Edited January 8, 2017 by Jason M

[bluearmy]

3 minutes ago, IDO said:

I don't know what you are on about, but a dew point of -15c is pretty unprecedented in recent times for De Bilt. In 2016 the min for Amsterdam was about -8c, ditto 2015:

  Source: here

If you believe that -15c Dew points are close to verifying, then I don't know what i can say! March 2013 was the previous time, -17c. The chart for that day very different:

To get to that extreme then the run must have gone off kilter sometime before, probably before D8. 

Sometimes when an algorithm churns out something as weird as that, you have to ignore it, otherwise all our experience is wasted. If the pattern was different with maybe amplification upstream, then this type of scenario would be slightly more plausible, but it isn't.

Anyway I do not think to point out that the likelihood of a dew point of -15c being implausible as high is "obtuse". 

 

It evolved in a similar fashion to the op. Nothing odd. Ridge a bit more suppressed south than the op but not enough to prevent a decent easterly flow to advect the low uppers west.