Model Output Discussion - mid Autumn

[IanT]

2 minutes ago, Nick F said:

Yes, think AI weather models, with its machine-learning hindsight approach (think it looks back at 40 years of data) will be an important addition along with NWP, with it's physics approach. We'll soon be looking at the AI run as well as the operational run I guess!

Google DeepMind’s AI Weather Forecaster Handily Beats a Global Standard | WIRED

WWW.WIRED.COM

---

Machine learning algorithms that digested decades of weather data were able to forecast 90 percent of atmospheric measures more accurately than Europe’s top weather center.

DeepMind AI accurately forecasts weather — on a desktop computer

WWW.NATURE.COM

---

The machine-learning model takes less than a minute to predict future weather worldwide more precisely than other approaches. The machine-learning model takes less than a minute to predict future weather worldwide...

 

[MATTWOLVES 3]

You beauty  

Met could.be getting anxious now.

Daily Express could be having babies

[terrier]

34 minutes ago, Met4Cast said:

No, but it will have taken into account tropical forcing & I suspect GLOSEA/EC decider models which evidently are not supportive of the det runs being shared here. 

The Met are the best in the business. 

Have to agree with you on this one. The Glosea/Ec must not be buying this. As you say the met are probably the best in the business and they certainly aren’t buying what the GFS is churning out at the moment. Guess something has to give in the next few days. One thing for sure one way  or another is going to have egg on its face. Until the met are fully onboard I won’t be rushing out to get the de ice just yet. 

[Rob 79812010]

11 minutes ago, andy989 said:

For you perhaps. I did very well, best December since 2010 for me. 

For me also. Good snow for Manchester I recall. Well, good for Manchester anyway

[steveinsussex]

Public service announcement- please nobody tell friends or family it may get cold with the potential for snow at some point

 

thank you 

[Chesil View]

Just now, Nick F said:

Yes, think AI weather models, with its machine-learning hindsight approach (think it looks back at 40 years of data) will be an important addition along with NWP, with it's physics approach. We'll soon be looking at the AI run as well as the operational run I guess!

It's an interesting concept Nick. The question has often been asked in the model thread do nwp models take note of history or have a 'memory' the answer always being no, because they simply take into account the current physics of the atmosphere.  Whereas a history based model will be able to take into a account that even though the physics says this we now that previous identical set ups have seen 2 in every 10 lows to track further south /north etc than the physics alone would suggest.

[*Stormforce~beka*]

Right Beka is getting interested now ... How many models are FOR snow V models against snow please? Or prehaps I should say snow potential ...

[booferking]

1 minute ago, Penicuikblizzard said:

Nothing secured at this stage could all still change, but on a positive note it’s heading in the right direction 

Yes as it meant to read it would be secured if that chart game to fruition.

[Mark wheeler]

Woahh . Is that snow or cold rain 

 

or neither

Edited November 22, 2023 by Mark wheeler

[Chesil View]

4 minutes ago, terrier said:

Have to agree with you on this one. The Glosea/Ec must not be buying this. As you say the met are probably the best in the business and they certainly aren’t buying what the GFS is churning out at the moment. Guess something has to give in the next few days. One thing for sure one way  or another is going to have egg on its face. Until the met are fully onboard I won’t be rushing out to get the de ice just yet. 

In reality the Met Office only jump on board when it's beyond reasonable doubt in other word 80% or more likelihood.

Today's update will not have taken in to account this afternoons increase in likelihood of extended cold across all models.

It may well change again for tomorrow's update.

Edited November 22, 2023 by Chesil View

[Ali1977]

Not super cold uppers at around -4 to -8c  but neither was last December but it was really cold at ground level - memorable spell coming up if this works out 

[Tim Bland]

Plenty of time to get this low a bit further north…

 

Edited November 22, 2023 by Tim Bland

[northwestsnow]

UKMO 🫨

 

 

 

[Bats32]

Just now, Chesil View said:

In reality the Met Office only jump on board when it's beyond reasonable doubt in other word 80% or more likelihood.

Today's update will not have taken in to account today's increase in likelihood of extended cold across all models.

It may we change for tomorrow's update.

The last thing they want to do is be quatied by all the click bait papers. They usually take the conservative route.

[Chesil View]

1 minute ago, Bats32 said:

The last thing they want to do is be quatied by all the click bait papers. They usually take the conservative route.

And rightly so.

[Mark wheeler]

Just now, northwestsnow said:

UKMO 🫨

 

 

 

TBH thats the one I always look for as others have said if UKMO says no in my experience it rarely happens . Nice to see.

[andy989]

I know people bash the UKMO for being late to parties, but having it on board, late or not, is important. And as for their extended outlook, they always take the safer route. Mentions of cold in their write ups only happen when they are very certain of it. It’s always been like this. 

Edited November 22, 2023 by andy989

[Metwatch]

2 minutes ago, Tim Bland said:

Plenty of time to get this low a bit further north…

So close yet so far, but the goods will be delivered one day, we hope!

 

[Ice Day]

We're now in the 'just for fun' stage of the run, but at 228 the deep cold to the northeast of Scandi could be about to come into play.  That really would be something!

 

[Ali1977]

More WAA on the way

[Stabilo19]

11 minutes ago, Mucka said:

So UKMO, Icon, GEM now all pretty much on board with the overall GFS progression out to 144, ECM odd one out but moved a little toward GFS this morning like UKMO.

Love to see it! Pretty high confidence in a cold spell of weather in our locale now. 

Things can change by next week but with such cross model support.. the weather is looking seasonal for a change! 

Edited November 22, 2023 by Stabilo19

[Catacol]

19 minutes ago, Nick F said:

Yes, think AI weather models, with its machine-learning hindsight approach (think it looks back at 40 years of data) will be an important addition along with NWP, with it's physics approach. We'll soon be looking at the AI run as well as the operational run I guess!

And here is the Google Graphcast AI forecast for Tuesday. Pick the bones out of this one - sitting right on top of the fence. By the end of the week it has the cold pushed north and the UK under a trough.

Link to ECMWF running an experimental page created by the Google AI is here:

Experimental: GraphCast ML model: Mean sea level pressure and 850 hPa wind speed

CHARTS.ECMWF.INT

---

**ECMWF is now running an experimental suite to evaluate the forecasting skill of a series of publicly available data-driven forecast models initialised by ECMWF 4D-Var analysis. ECMWF provides the output for...

 

Edited November 22, 2023 by Catacol

[Speedbird]

1 minute ago, Metwatch said:

So close yet so far, but the goods will be delivered one day, we hope!

 

After the recent floods, Aberdeenshire about to get buried . . .

[Lukesluckybunch]

5 minutes ago, northwestsnow said:

UKMO 🫨

 

 

 

I love this chart,one of the best.-8 uppers there I reckon!

[January Snowstorm]

Charts settling on a colder option, if not even wintry! Equally it is right to say that the gfs has backed away from its raging Easterly and now shows a messier and slower path to cold. Still brilliant, but worth noting all the same

Slower as in Wednesday now rather than Tuesday

Edited November 22, 2023 by January Snowstorm