Model Output Discussion - mid Autumn

[Met4Cast]

Just now, Mike Poole said:

GFS 6z at T324?  Really?

I think you're very much missing the point there lol

[throwoff]

6 minutes ago, CoventryWeather said:

 

Going back to what's likely, I'm still not convinced the models have found a solution 

I make you right here. If you took the current runs across the models as gospel we are about to see widespread snow with low temperatures at the same time as mild dry weather.

We say it every year, part of snow is nowcasting, I have to say from years of following the models every winter we seem to do better from events that are still nowhere near settled 4 days out than the +240 wonder charts that never come to fruition. 

Maybe I am leaning into the over confident early but that the models still cant make up their mind for Friday is a good thing in my experience.

[CryoraptorA303]

2 minutes ago, throwoff said:

I make you right here. If you took the current runs across the models as gospel we are about to see widespread snow with low temperatures at the same time as mild dry weather.

We say it every year, part of snow is nowcasting, I have to say from years of following the models every winter we seem to do better from events that are still nowhere near settled 4 days out than the +240 wonder charts that never come to fruition. 

Maybe I am leaning into the over confident early but that the models still cant make up their mind for Friday is a good thing in my experience.

Hard agree, the events that actually materialise seem to spawn in from nowhere only a few days before they occur. Vs the long-range predicted events never materialising e.g. the hyperforecasted second blast after last December. First it was late December, then it was new years, then it was mid-January, then it was early February, then by chance a colder pattern materialised in early March but it was too late by then.

[Mike Poole]

12 minutes ago, Met4Cast said:

I think you're very much missing the point there lol

Deliberately so .

Don’t get me wrong, I’m not in disagreement with what you have been saying about the background signals not being supportive of amplification, all indications are the MJO is moving away from being helpful and AAM is falling.  It is the very over-confident tone to your posts that slightly grates with me, because there are potentially other unknowns in the equation, such as the behaviour of the strat, SSTs in the Atlantic, but also globally which this year are out of any recent precedent, so the potential for wildcards deviating the evolution from what is expected should at least be acknowledged.  

Forecasts based on teleconnections didn’t go that well in the summer, this year.  Mind you, neither did the models in the medium range, to be honest.

Edited November 20, 2023 by Mike Poole

[TSNWK]

Jeeze.... I'm not sure what I want more if GFS reverted to its previous runs from tonight

Snow and ice days potential or a kick in the nuts for smug posting     

it's the weather and far more important things to get into twist over...

[Dennis]

friday 24th

 

[TSNWK]

1 minute ago, Mike Poole said:

Deliberately so .

Don’t get me wrong, I’m not in disagreement with what you have been saying about the background signals not being supportive of amplification, all indications are the MJO is moving away from being helpful and AAM is falling.  It is the very over-confident tones to your posts that slightly grates with me, because there are potentially other unknowns in the equation, such as the behaviour of the strat, SSTs in the Atlantic, but also globally which this year are out of any recent precedent, so the potential for wildcards deviating the evolution from what is expected should at least be acknowledged.  

Forecasts based on teleconnections didn’t go that well in the summer, this year.  Mind you, neither did the models in the medium range, to be honest.

Agree on teleconnections having challenges during the summer.. I remain in the to be convinced category on them   i still Currently think there is too much chaos theory to have confidence beyond educated theories beyond about 4 weeks  forecasting 

[Met4Cast]

Just now, Mike Poole said:

Deliberately so .

Don’t get me wrong, I’m not in disagreement with what you have been saying about the background signals not being supportive of amplification, all indications are the MJO is moving away from being helpful and AAM is falling.  It is the very over-confident tones to your posts that slightly grates with me, because there are potentially other unknowns in the equation, such as the behaviour of the strat, SSTs in the Atlantic, and also elsewhere which this year are out of any recent precedent, so the potential for wildcards deviating the evolution from what is expected should at least be acknowledged.  

Forecasts based on teleconnections didn’t go that well in the summer, this year.  Mind you, neither did the models in the medium range, to be honest.

There is no "tone" to my posts, they are text based, it's impossible to ascertain tone. The tone you are reading my posts is entirely in your own head, I'm afraid. 

The point of my post was to highlight how when NWP correlates to what teleconnections are telling us then generally we can have a higher level of confidence in the NWP output, I was using the 06z GFS as an example of that to show how that run in particular is congruent with the background signals vs how the recent Greenland High scenario wasn't. 

You're absolutely right that teleconnections failed during summer, AAM collapsed during July and never really recovered, this caught practically everyone on the back foot. As we know though, moving into the winter period is when things become broadly more predictable because more primary drivers come into play. All we can do is put forward our best educated guess on how the pattern will/may evolve and then sit back and analyse the situation, evolving the forecast if necessary on the back of new evidence, that's all I'm doing. 

I fully suspect that if I was suggesting bitter cold easterlies & blizzards were on the way less people would be taking issue with my posts, but that's the nature of this thread unfortunately.

[Mike Poole]

Just now, Met4Cast said:

I fully suspect that if I was suggesting bitter cold easterlies & blizzards were on the way less people would be taking issue with my posts, but that's the nature of this thread unfortunately.

That - is entirely possible!

[TSNWK]

2 minutes ago, Met4Cast said:

There is no "tone" to my posts, they are text based, it's impossible to ascertain tone. The tone you are reading my posts is entirely in your own head, I'm afraid. 

The point of my post was to highlight how when NWP correlates to what teleconnections are telling us then generally we can have a higher level of confidence in the NWP output, I was using the 06z GFS as an example of that to show how that run in particular is congruent with the background signals vs how the recent Greenland High scenario wasn't. 

You're absolutely right that teleconnections failed during summer, AAM collapsed during July and never really recovered, this caught practically everyone on the back foot. As we know though, moving into the winter period is when things become broadly more predictable because more primary drivers come into play. All we can do is put forward our best educated guess on how the pattern will/may evolve and then sit back and analyse the situation, evolving the forecast if necessary on the back of new evidence, that's all I'm doing. 

I fully suspect that if I was suggesting bitter cold easterlies & blizzards were on the way less people would be taking issue with my posts, but that's the nature of this thread unfortunately.

You seem knowledgeable and clearly know your onions... have you thought about slipping in some emojis within your posts.  It's exactly why they were created initially.. add a little nauce to the written word

[Met4Cast]

1 minute ago, TSNWK said:

You seem knowledgeable and clearly know your onions... have you thought about slipping in some emojis within your posts.  It's exactly why they were created initially.. add a little nauce to the written word

Yes, perhaps that’s a good idea!  

Certainly not meaning to convey any kind of tone within my posts! 

[throwoff]

3 minutes ago, Mike Poole said:

That - is entirely possible!

The 'problem' with this thread is that is does pull out a cross section from ardent rampers to doom mongers. 

Sadly when the models are complex toys immediately go flying from multiple prams and the unstoppable force meets the immovable object.

FWIW the mods do a superb job, especially as the thread gets busier and busier.

[MattH]

Let's strip this all back shall we and highlight the key points;

1. Will it turn colder end of the week onwards? - Yes for many areas, the exception perhaps being far W and SW of the British Isles
2. Is this a resolute, text-book Greenland block? - No, it isn't, it is an amplification of the N Atlantic pattern (Atlantic Ridge regime) as expected for several days
3. Did the GFS 'win this one'? - Yes, IMO, it did. The consistency over the colder flow from the N, has been consistent from the GFS in some shape or form, EC poorest with the handling of this.
4. Is widespread snow likely by the weekend? - No, the pattern is likely to be more anticyclonic than cyclonic, but still some 'give and take' in that
5. Which areas could see snow? - Scotland and E England, primarily over higher ground, with little or no snow elsewhere
6. Is this the first 'noteworthy' cold spell of the season for many? - Yes, it is likely to be so

Without dredging up individual model runs and details, there is now high confidence for this cold spell to arrive as a cold front moves southwards on Thu into Fri. Logic dictates it will be in the N and E where the coldest air will be, less cold in the S and W. The 06Z GEFS highlights this now almost 'clean' transition and change in air mass, also worth noting that the 06z GFS and Control on the milder side into next week too.

The long-term still carries a trend for a return to more Atlantic-driven weather, but the signal for the eastward propagation of the tropical signal and MJO is gaining traction for December and a potentially very interesting second half of December may still be on the cards.

Despite some evidence for a strat and trop connection as well, there remains little evidence in NWP for an organised vortex either with certainly some 'unusual' goings on at the moment, in keeping with recent global extremes/variations/unknowns in past seasons too. 

Regards, Matt.

[ANYWEATHER]

Well just typical for the Gfs 06z, it’s now undoing the buttons as regards the phantom cold spell in true Gfs Style.

[throwoff]

2 minutes ago, MattH said:

 

The long-term still carries a trend for a return to more Atlantic-driven weather, but the signal for the eastward propagation of the tropical signal and MJO is gaining traction for December and a potentially very interesting second half of December may still be on the cards.

 

 

Matt, do you buy the 'get the cold in' arguement? In that setting up a decent cold spell early often leads to further interesting patterns as the season progresses? 

 

I realise we are looking more at statistically analysis than forecasting in that argument but from my understanding it tends to hold up.

[Dennis]

im very hopefull on winter chances 

lets wait details later this week 

[knocker]

37 minutes ago, throwoff said:

The 'problem' with this thread is that is does pull out a cross section from ardent rampers to doom mongers. 

Sadly when the models are complex toys immediately go flying from multiple prams and the unstoppable force meets the immovable object.

FWIW the mods do a superb job, especially as the thread gets busier and busier.

There is a strong rumour doing the rounds that this thread is going to be renamed the gymkhana thread.

[adamgooner]

1 hour ago, Met4Cast said:

Just to go back to this post here & how teleconnections often lead NWP outputs, the 06z GFS is picking up this idea in the extended and running with it for early December.

Due to the drivers mentioned in the quoted post it means we can have more confidence in this evolution (details aside) from the GFS because it's well founded & advertised within the broader drivers, vs the recent Greenland high runs we've been seeing which haven't been supported. 

NWP doesn't lead, teleconnections do and that's why it's important to look at the entire picture when looking ahead vs just following individual det runs.

How often do you find a chart circa 300 hrs plus come to fruition?

This is cherry picking and misleading for learners. 
 

The models have been flipping back and forth of late inc the much lorded ecm. 
 

There is no overwhelming evidence across the suite to support the chart you’ve posted above, unless you pay high regard to charts deep into FI which rarely materialise.

[Redbull165]

In terms of accuracy usually the GFS00Z, 12Z runs are better than 6z,18z.

[throwoff]

Just now, Redbull165 said:

In terms of accuracy usually the GFS00Z, 12Z runs are better than 6z,18z.

the 18z is the worst in my experience, it isn't called the pub run for nothing.

[Met4Cast]

12 minutes ago, adamgooner said:

How often do you find a chart circa 300 hrs plus come to fruition?

This is cherry picking and misleading for learners. 
 

The models have been flipping back and forth of late inc the much lorded ecm. 
 

There is no overwhelming evidence across the suite to support the chart you’ve posted above, unless you pay high regard to charts deep into FI which rarely materialise.

I think you need to go back & read what I actually said, I wasn’t saying that specific chart would come to frution exactly as shown, I was using it to illustrate the point re: teleconnective signals & NWP modelling and why we can have more (or less in some cases) confidence in certain situations, such as the one so highlighted in the post you’re responding to  

When teleconnections point left but NWP points right, there’s a disconnect & confidence plummets, when both are pointing in the same direction confidence improves. 

Edited November 20, 2023 by Met4Cast

[Redbull165]

6 minutes ago, throwoff said:

the 18z is the worst in my experience, it isn't called the pub run for nothing.

I can show you also for other models. In the last few weeks ECMWF was the most accurate model, followed by UKMO, GFS. But in the last days GFS is surprisingly the best one. The point is, that every model has its flaws.

 

Edited November 20, 2023 by Redbull165

[adamgooner]

1 minute ago, Met4Cast said:

I think you need to go back & read what I actually said, I wasn’t saying that specific chart would come to frution exactly as shown, I was using it to illustrate the point re: teleconnective signals & NWP modelling and why we can have more (or less in some cases) confidence in certain situations, such as the one so highlighted in the post you’re responding to  

But it’s an isolated chart at 300 plus hours - surely it doesn’t add any value to forecasting what might happen up top or on the ground.

[Met4Cast]

Just now, adamgooner said:

But it’s an isolated chart at 300 plus hours - surely it doesn’t add any value to forecasting what might happen up top or on the ground.

It adds value to the post I made & provides context to what I was saying. No where in that post did I say that chart would come off exactly as is, indeed, I even caveated the chart with “(details aside)”

Would you be taking this issue if it was a cold blocked chart I had shown? I doubt it

[Metwatch]

Been a few conundrums on here today by the looks of things.

To me there's still too much emphasis on single deterministic runs.

The GFS and ECM ensembles for here in the Midlands are still decent for a consistently chilly airmass to be here and so over most of the country, the ensemble means are below the 30 year average all way up to the end of this month. The GFS 06z operational going for a milder incursion early next week, but a look at the ensemble graph, and as of this morning, it isn't very well supported with the rest of the suite. That may well change, but at the moment it's not fully known. The ECM is less keen on that. Later next week we might continue to have a chilly airmass or become even colder, it's a total guess at the moment, my feeling is as we start December more likely than not a westerly zonal regime does become established again though.

Yes it was quite unlikely to be as cold as some other det runs showed from the GFS with those frontal snow events, but the ECM hasn't fully hit the nail on the head either. To me the GFS picked up on the general pattern with the mid Atlantic ridge then a longer fetched northerly a bit better compared to the ECM.

Will we all get buried in snow like late November 2010 and have harsh frosts courtesy of a picture perfect Greenland high? Of course not, but it's still going to be the first decent cold spell of the season. Any snow of course favoured for northern / north eastern hills as is often the case with these setups in late autumn.

 

 

Edited November 20, 2023 by Metwatch