Model discussion: Hunt for cold - The start of 2022

[Ali1977]

2 minutes ago, northwestsnow said:

Better mean.

Need to see the momentum continue tomorrow..

Yep, we are still flapping about without any clear direction which I guess is quite normal 10 days out.  The ECM op route to cold was quite complex, it would be nice to see some straight forward ridging. That cluster 1 Mike P posted beyond T264 is a belter mind you

[Kasim Awan]

12 minutes ago, chris55 said:

Not in the context you quoted me with though. 

Any day 10 chart will lose reliability, but not every day 10 chart loses resolution.

I think GFS reduces resolution at day 10 onwards. Hence the term FI. 

Important to clarify. 

 
 

 

As reliability drops so does the resolution of the spread of possibilities and therefore the resolution of any judgement / forecast that can be made, this is why I originally said resolution, but yes perhaps I should have rephrased this. There is pixel resolution which I think is what you are referring to, which is also different from macro scale modelling resolution - i.e. the complexity of features will drop off post day 7 on most output. The term FL is related to reliability, not pixel or macro complexity resolution. A chart can be 100% correct but pixel / macro complexity resolution poor - e.g. turning down the resolution of youtube video will still be the same video.

Edited January 8, 2022 by Kasim Awan

[Mike Poole]

3 minutes ago, Ali1977 said:

Yep, we are still flapping about without any clear direction which I guess is quite normal 10 days out.  The ECM op route to cold was quite complex, it would be nice to see some straight forward ridging. That cluster 1 Mike P posted beyond T264 is a belter mind you

Also, in case you hadn’t spotted it, nerdy point here!, the representative perturbation chosen to illustrate the cluster was ‘perturbation 0’ - which is the control.  (Op run obviously finishes at T240.)

Edited January 8, 2022 by Mike Poole

[geordiekev]

Think I must have been doing all this model reading wrong for the last decade from reading last few pages.

So at day 14 all models are laughable & if trend is still there at day 10 it's impossible. Really should pin all your advice for future.

Thanks for the tips all

Edited January 8, 2022 by geordiekev

[northwestsnow]

5 minutes ago, Kasim Awan said:

As reliability drops so does the resolution of any judgement / forecast that can be made, this is why I originally said resolution. That is different from pixel resolution which I think is what you are referring to, which is also different from macro scale modelling resolution - i.e. the complexity of features will drop off post day 7 on most output. The term FL is related to reliability, not pixel or macro complexity resolution. A chart can be 100% correct but pixel / macro complexity resolution poor.

TBH I'm just delighted I've some interest .

It may come to nothing but a subtle change in text by Exeter today, and an improved EC and attendant clusters suggests Jan is not completely lost.( I will be first to openly admit I was already feeling despondent about Jan's prospects).

Edited January 8, 2022 by northwestsnow

[carinthian]

On 07/01/2022 at 16:55, carinthian said:

 Getting there slowly. Heck its like drawing teeth to tap into that cold. Some real cold air on this chart heading out of the Arctic towards The British Isles. Would just prefer that relocation of the TPV a bit further south with its core over The White Sea to give some extra kick. More important lying development is the blocking out in the North Atlantic. Still early days but sometimes a slow trigger low can produce the best results.

C

Models continue to struggle out at 10 days. The chart below from ECM  nows shows a trigger Low that would sink some much colder air along an increasingly tight thermal zone, similar to what was shown last evening on GFS.  So still some hope for snow in the mix for you guys looking at the day 10 charts currently on offer. The ECM current op offers the best chance against the other models this evening. Probably change again in the morning !

 C

Edited January 8, 2022 by carinthian

[damianslaw]

Reading the models today, looks like we will see high pressure building from the south attempt to ridge north but then sinks SW into mid atlantic eventually as we see formation of a scandi trough. This is the direction of travel I'm seeing. Question is how far north and west may the ridge end up, could elongate N-S and allow a northerly plunge as suggested by ECM and GFS 10 day range, or it may be closer and we see chilly NW airflow riding in over the top. Either way, a fairly average period coming up temperature wise, with potential for something colder around 17th January onwards.

[Kasim Awan]

7 minutes ago, northwestsnow said:

TBH I'm just delighted I've some interest .

It may come to nothing but a subtle change in text by Exeter today, and an improved EC and attendant clusters suggests Jan is not completely lost.( I will be first to openly admit I was already feeling despondent about Jan's prospects).

Yes, we all know in here this winter requires extra caution compared to normal due to the tendency for modelling to underplay the unusually strong Azores high within their extended ranges (post 200h). This is why all the good output is generally 200h+, imo, as modelling only has "now"/observational data for their reference of the strong Azores, and this is lost over time in a run which results in a weakened Azores high & good output as a result. Not 100% but I think a bit of this is going on imho. So we need to be very cautios this year. There are reasons for this inflated Azores high - the telleconnection experts amongst us will know

Edited January 8, 2022 by Kasim Awan

[northwestsnow]

1 minute ago, Kasim Awan said:

Yes, we all know in here this winter requires extra caution compared to normal due to the tendency for modelling to underplay the unusually strong Azores high within their extended ranges (post 200h). This is why all the good output is generally 200h+, imo, as modelling only has "now"/observational data for their reference of the strong Azores, and this is lost over time in a run which results in a weakened Azores high. Not 100% but I think a bit of this is going on imho. So we need to be very cautios this year. There are reasons for this inflated Azores high - the telleconnection experts amongst us will know

 

MOGREPS I believe has trended colder ...

I've been working but I'm pretty sure the above is the case...certainly next week...

[Kasim Awan]

6 minutes ago, northwestsnow said:

 

MOGREPS I believe has trended colder ...

I've been working but I'm pretty sure the above is the case...certainly next week...

Yeah, there is interest but when we've got an ecm mean like this, the interest is almost no more than the usual cluster of cold outliers found at any point in the winter.

I remember in early January 2016, there was so much "interest" as models underestimated the Azores in their extended range. So many promising ecm day 10 charts resulted from this underestimation. As the days ticked by the models got a better grip on the strong vortex signal and the promising charts were lost. I'd say the same is happening this winter.

Edited January 8, 2022 by Kasim Awan

[geordiekev]

10 minutes ago, Kasim Awan said:

Yes, we all know in here this winter requires extra caution compared to normal due to the tendency for modelling to underplay the unusually strong Azores high within their extended ranges (post 200h). This is why all the good output is generally 200h+, imo, as modelling only has "now"/observational data for their reference of the strong Azores, and this is lost over time in a run which results in a weakened Azores high & good output as a result. Not 100% but I think a bit of this is going on imho. So we need to be very cautios this year. There are reasons for this inflated Azores high - the telleconnection experts amongst us will know

Could be vice versa as the Christmas prospects which was modelled quite far in advance underplayed the Azores which eventually stretched too high North, ending high up through Greenland and into Arctic as it came closer, so ended a smidge too high for our small Island. Or is that what you're suggesting & I've misread 

Edited January 8, 2022 by geordiekev

[Don]

2 minutes ago, Kasim Awan said:

Yeah, there is interest but when we've got an ecm mean like this, the interest is almost no more than the usual cluster of cold outliers found at any point in the winter.

I remember in early January 2016, there was so much "interest" as models underestimated the Azores in their extended range. So many promising ecm day 10 charts resulted from this underestimation. 

That mean looks somewhat meh! 

[BLAST FROM THE PAST]

1 hour ago, CreweCold said:

Yeah. D10 ECM is about as useful as a concrete parachute, same with most model op runs at that timeframe to be honest.

We would have had some decent cold spells over the last few years had those D10 ECM charts come off, that's for sure.

Unless the run up to it is pretty plausible?  When that happens day 10 isn’t so outlandish?   

 

BFTP

17 minutes ago, carinthian said:

Models continue to struggle out at 10 days. The chart below from ECM  nows shows a trigger Low that would sink some much colder air along an increasingly tight thermal zone, similar to what was shown last evening on GFS.  So still some hope for snow in the mix for you guys looking at the day 10 charts currently on offer. The ECM current op offers the best chance against the other models this evening. Probably change again in the morning !

 C

I think it’s coming, but maybe still too soon, but a trigger imo is coming….this month

BFTP

[Kasim Awan]

12 minutes ago, geordiekev said:

Could be vice versa as the Christmas prospects which was modelled quite far in advance underplayed the Azores which eventually stretched too high North, ending high up through Greenland and into Arctic as it came closer, so ended a smidge too high for our small Island. Or is that what you're suggesting & I've misread 

Yes, the models only had observational data to work with for a computational signal for strength of the Azores high. Once this signal is lost as the run progresses towards the extended range, incorrectly weak resolution of the Azores high ensues, resulting in the overcooked cold patterns seen throughout this winter. Happened first over Christmas in a big way, so it could happen again. I'm not saying there is no interest atm, just saying caution is needed (more than normal). This is my new approach to interpreting the output, based on the behaviour of modelling so far this winter.

Edited January 8, 2022 by Kasim Awan

[Mike Poole]

So where are we right now?  I’m not staying up for the pub run, early start tomorrow!  Well for a week or so now this place has been a strange place with those of a cold persuasion (I think it is fair to say) looking fairly long range for thin gruel, and of course cherry picking fruity ensemble members, while those who are convinced there is no cold to come at all in January (I think it is also fair to say) saying so repeatedly without backing it up with any evidence at all (I exempt @Catacol and @CreweCold from that generalisation because they have provided their reasoning).   But I sense a shift based on the model output today:

• The evolution after the reliable relys heavily on where the bulk of the trop PV ends up, and while there is definitely an early movement to the east, whether it all stays there, how much remains as a Canadian lobe has been massively open to question.
• The 12z suite, and ECM clusters, seem to be reinforcing the idea that the shift of the PV to the east will be more substantial, allowing an Atlantic ridge to form and the Azores high to retract, thereby allowing a trough into Southern Europe. 
• For a longer term cold pattern, this is now a 40% chance at most (going off the 12z clusters 20/51 - that could be a flash in the pan, need to see it repeated on next couple of runs).
• For shorter term northerlies (topplers), I would say there is a better chance than 40% (75%?) we will see something the week after next.  
• Today, with the ECM 12z we now have something to work with within 10 days.
• There is still massive uncertainty about how the next 2 weeks pan out, and anyone saying they know now are talking out of their backsides.  
• I think there will be some twists and turns to come if we are to get this down to T0!

Best regards

Mike

Edited January 8, 2022 by Mike Poole

[Kasim Awan]

Just making it clear that not at any point have I said that there's no chance this month. The chance is much lower than normal though - there are obvious reasons for this, explanation is enough, the evidence is clear..

[E17boy]

28 minutes ago, Kasim Awan said:

Yeah, there is interest but when we've got an ecm mean like this, the interest is almost no more than the usual cluster of cold outliers found at any point in the winter.

I remember in early January 2016, there was so much "interest" as models underestimated the Azores in their extended range. So many promising ecm day 10 charts resulted from this underestimation. As the days ticked by the models got a better grip on the strong vortex signal and the promising charts were lost. I'd say the same is happening this winter.

So true as I have mentioned in previous posts sometimes it’s like the Bermuda Triangle tasty charts that are showing in the 10 day timeframe seem to mysteriously disappear, alas that’s the luck we sometimes have being on this island. Hopefully we will get there one day I wish it’s sooner than later, because come February if nothing has materialised it won’t only be the models we will be looking at in agony but the clock as well. Hopefully it won’t come down to that final straw

[geordiekev]

Although I don't have the Northern Hemisphere view from 7 days ago what we're now seeing on ECM was pretty well forecast by GFS & ECM certainly took some time to catch up.

I've been touting the seaboard shortwave for the 11th being a bit of game changer bc this was what GFS had whilst others hadn't & helps in squeezing the Scandi High Pressure back across to the West of us. For me the GFS has been quite decent this last week as the 2 GFS images from the 2nd Jan show compared to tonights ECM for the same date

Edited January 8, 2022 by geordiekev
Typo

[Kasim Awan]

3 minutes ago, E17boy said:

So true as I have mentioned in previous posts sometimes it’s like the Bermuda Triangle tasty charts that are showing in the 10 day timeframe seem to mysteriously disappear, alas that’s the luck we sometimes have being on this island. Hopefully we will get there one day I wish it’s sooner than later, because come February if nothing has materialised it won’t only be the models we will be looking at in agony but the clock as well. Hopefully it won’t come down to that final straw

Yep. Not enough in it atm to be confident that the persistent Azores slug will be broken. It's a persistent and highly influential feature (Azores high / hadley cell) vs a timid, relatively weak and inconsistent source (mid latitude blocking). The latter is no contender atm, though a laxing of the Azores high will give the later a chance to take hold (20%), however this usually results in topplers which is the synoptic outcome of a combined cold signal + Azores signal.

February can bring a change of fortunes following a mild Dec / Jan - 2017/18 an example. However, it's not the statistically favoured outcome.

[chris55]

1 hour ago, Kasim Awan said:

As reliability drops so does the resolution of the spread of possibilities and therefore the resolution of any judgement / forecast that can be made, this is why I originally said resolution, but yes perhaps I should have rephrased this. There is pixel resolution which I think is what you are referring to, which is also different from macro scale modelling resolution - i.e. the complexity of features will drop off post day 7 on most output. The term FL is related to reliability, not pixel or macro complexity resolution. A chart can be 100% correct but pixel / macro complexity resolution poor - e.g. turning down the resolution of youtube video will still be the same video.

What?

Model resolution is very different to YouTube resolution! 
 

 

Imagine running ECM at 30km resolution (like a grainy YouTube video) 

Currently ECM is run at 9km and the ensembles at 18km.

You referred to ECM 240 losing resolution, it doesn’t .

 

[Don]

23 minutes ago, Kasim Awan said:

Yep. Not enough in it atm to be confident that the persistent Azores slug will be broken. It's a persistent and highly influential feature (Azores high / hadley cell) vs a timid, relatively weak and inconsistent source (mid latitude blocking). The latter is no contender atm, though a laxing of the Azores high will give the later a chance to take hold (20%), however this usually results in topplers which is the synoptic outcome of a combined cold signal + Azores signal.

February can bring a change of fortunes following a mild Dec / Jan - 2017/18 an example. However, it's not the statistically favoured outcome.

December 2017 wasn't mild though at 4.8C, featured a few cold/snowy snaps and January 2018 was a cold month for northern UK.  I don't think we will see a BFTE this year like in February/March 2018, but hopefully at least some transient snow events including southern regions, most likely in February.

Edited January 8, 2022 by Don

[Kasim Awan]

1 minute ago, chris55 said:

What?

Model resolution is very different to YouTube resolution! 
 

 

Imagine running ECM at 30km resolution (like a grainy YouTube video) 

Currently ECM is run at 9km and the ensembles at 18km.

You referred to ECM 240 losing resolution, it doesn’t .

 

Eh? What do you think the 30km value means? It's the spatial esolution of the raw data outputted by the model. Much like the pixels of a screen which also have specific values. That is a fact, no matter how the data is generated. 

[chris55]

59 minutes ago, Kasim Awan said:

Yes, the models only had observational data to work with for a computational signal for strength of the Azores high. Once this signal is lost as the run progresses towards the extended range, incorrectly weak resolution of the Azores high ensues, resulting in the overcooked cold patterns seen throughout this winter. Happened first over Christmas in a big way, so it could happen again. I'm not saying there is no interest atm, just saying caution is needed (more than normal). This is my new approach to interpreting the output, based on the behaviour of modelling so far this winter.

Surely this applies to all aspects of the data and not just the Azores high? 
 

Obsevational = starting data

Computational = Model running and computing an outcome.
 

I’m very confused by your reasoning TBH. 

[Kasim Awan]

Just now, chris55 said:

Surely this applies to all aspects of the data and not just the Azores high?

Not neccessarily. If a feature is in a highly anomalously unusual state caused by a telleconnection which the model doesn't have great linkage with, then observational data is the only reference for the abnormality.

[chris55]

2 minutes ago, Kasim Awan said:

Eh? What do you think the 30km value means? It's the spatial esolution of the raw data outputted by the model. Much like the pixels of a screen which also have specific values. That is a fact, no matter how the data is generated. 

Yes but the resolution of 9km refers to the “clarity” of the data inputted into the starting data. 
 

if you input 30km grid data you get a less detailed and accurate forecast. Like a poor quality YouTube video. (For example)