Model output discussion - the beast arrives

[Summer Sun]

Hot off the press - UKMO extended keeps the easterly feed going

[chris55]

3 minutes ago, Snipper said:

Bearing in mind how rubbish the BBC weather forecast and app has been in recent times who has complained?

If yes what has been the best point of contact?I

It annoys me they keep quoting the Met Iffice when they have sacked them. 

They must still use morgeps and UKMO as Ian specifically quoted 'met office models" on his 650pm forecast.

what kind of contracts they have negotiated with meto group and the Met Office is not for us mere mortals to know, but it seems likely they have access the whole suit of models.

sorry mods off topic I know.

 

[Frosty.]

10 minutes ago, Summer Sun said:

Hot off the press - UKMO extended keeps the easterly feed going

That's bizarre but brilliant at the same time!! I guess we now need to see the Gfs changing significantly since the 12z shows 12 / 13c for the south on sunday!!

Edited February 26, 2018 by Frosty.

[DOdo]

Not convinced! with the  models  Just going enjoy thiis Cold Snap.and the snow this will bring

[Frosty.]

23 minutes ago, Frosty. said:

That's bizarre but brilliant at the same time!! I guess we now need to see the Gfs changing significantly since the 12z shows 12 / 13c for the south on sunday!!

Although the Gfs 12z op shows a very mild sunday for the s / se..the GEFS 12z mean is quite a lot colder than that!

Edited February 26, 2018 by Frosty.

[Catacol]

48 minutes ago, Quicksilver1989 said:

I think by then people will have had enough tbh and will lose interest in a half hearted cold spell. If we keep the continental air and Friday becomes a major snow event I think the beast will be seen as living up to the hype and we could have more snow events and keep the notable cold going. However the GFS and ECM operationals are pretty disappointing and if it gets whisked away I'd be happy to see the back of any cold weather until later on in the year.

Still hanging in the balance though but even UKMO shows milder air getting ahead of the low on Friday morning for a breif time. If the 00zs operationals tomorrow fail to stop the mild air advance by saturday I think the rest of the ensembles may follow as we will be getting close to the event. In which case people away from the eastern areas will probably rate December's cold snap higheras that at least didn't see the build up of snow melt quickly.

Agreed. There’s a lot of excitement on the news for the north and east, but for central southern England and the south-west there is the distinct chance that little to no snow will fall by Thursday, and then if mild air pushes north with any speed the spell quickly becomes a snap and one memorable for a blizzard that turns to rain and takes any enjoyment away with it.

But BA is also right - we don’t know yet what Fri/Sat will bring. There is time for improvements and as we all know 100miles south with a touch more resistance from the block could turn the whole event into solid snow for most. I feel thoroughly sorry for anyone that might miss out this week, because I know myself how hard it is going to be to take failure if it comes. Until Saturday this setup looked fantastic - it is now on a knife edge no matter how cold the next 2 days are going to be, away from the north and east at least.

So as expected no answer tonight. I never expected one. I think this will run until the 00z tomorrow night. By then at 24-36 hours range we have to think high res will have a handle on the evolution.

I’ll enjoy the cold air tomorrow, and the same the day after, but the sudden marginality for those in the S/SW after Thursday has taken quite a bit of the excitement away. For those in the north and east - enjoy what comes your way. Currently, with the obvious caveat that this could still turn in favour of all coldies everywhere, I envy you.

[Singularity]

6 or so hours of heavy freezing rain followed by a couple of heavy snow?

That from ECM for CS and eastern parts of SW England is a recipe for disaster. Hopefully the low will behave more like ICON goes with and we’ll have a relatively less severe blizzard.

[Mike Poole]

I think it's worth posting in some detail, the ECM 12z ensemble mean charts. Let's take the 850 temps first, here from T72 on:

Not above 0C anywhere.  Now the pressure/heights

Swing back out west here (T92) but then

Looks like prolonging the cold to me.

[Catacol]

17 hours ago, Tracy Flick said:

NWP is traditionally purely Newtonian physics (no historical data use), but some companies are trying Artificial Intelligence ('Neural Networks', the self-learning way).

How I see these 2 approaches:

Using pure Newtonian physics (which is the traditional MetO way):  In the case of fluid dynamics, the balance of forces and the conservation of momentum, energy and mass are expressed through Navier-Stokes partial differential equations (mass, energy and momentum are conserved in everything, even car crashes... read about Newton's Laws for more).  But these NS equations are difficult to 'solve' both numerically and analytically... in fact we don't even know how to generally prove that physical solutions exist to them at all.  The best we can do is therefore create a simplified version of the NS equations and then use computational techniques such as 'finite difference methods' (see Wikipedia) to solve the many partial differential equations simultaneously: these iterate the evolution over increments of time and space (the space increments are the grid spacings that are often referred to.)  The starting point (t=0) is observed data combined with reanalysis data.  Obviously smaller increments of time and space give better results but are more computationally expensive.

These Newtonian NWP models should in theory be fine with unusual atmospheric states because the laws of physics are still always true... however, the approximations introduced by the physicists and programmers may mean that accuracy is only guaranteed for certain ranges of atmospheric states.  By making observations of unusual states we can test the validity of these approximations and improve them if necessary.

An example: If there's an experimental observation such as eg 'high sea surface temperatures in an ocean leads to warmer summers' , this will 1st be checked for good statistically correlation.  Then this will be checked on the computer model to see if the simulation output matches the experimental observation.  If it does, great, but if not it means something is either lacking from the code or something is in error in the code.  Then the possible sources of error are identified eg maybe evaporation needs to be modelled less simplistically, so they will try that.  This will mean increasing the complexity of the approximate partial differential equations, or adding new ones.

Nonlinear systems (eg the atmosphere) require exponential increases in computing power for linear increases in forecasting accuracy... I suspect there is therefore probably no real desire anymore to simply improve results by buying better computers because the forecasting improvement is increasingly tiny for each extra dollar spent.  It is surely instead more efficient to improve the formulation of the differential equations that have to be simultaneously solved... a question that could be asked is eg "how can I better model turbulence on the edges on tropical storms?" ... answering this question will lead to modifications of the set of partial differential equations.

GFS I think uses finite difference methods to solve the partial diff equations but they want to advance to finite volume methods.  ECMWF use spectral methods... these involve writing the solution as a sum of basis functions as is done in Fourier analysis.

 

The AI Neural Network way: 

Some private companies are it seems now trying to somewhat blindly get forecast results using Artificial Intelligence.  See https://blogs.microsoft.com/ai/hows-the-weather-using-artificial-intelligence-for-better-answers/

Basically this method I think (or at least one possible implementation) uses historic observations with known outcomes to 'train' the software.  Then given enough training it should be able spot patterns and give a correct output (the forecast).

This is similar to spectral methods in that we are trying to find weights to give to each element of a set of functions... BUT this time the desired weighting on a function is evolved towards using nothing but historic observations as a guide, whereas the ECMWF find the weights by 'simply' plugging the sets of functions into the partial differential equations.

For an introduction to weather with neural networks and practising using it yourself:   https://www.amazon.co.uk/Neural-Network-Programming-Java-Souza/dp/178588090X

As computing power increases AI might become more important in forecasting, BUT weather has sensitivity to initial conditions so perfect AI forecasting can surely only be achieved with an infinite amount of training data(!)... that's to say AI methods must have at least the same fundamental limitation as classical physics methods.  I suspect that it is with unusual atmospheric states that AI will seriously struggle.

Maybe combining AI and traditional NWP in some way is a way some might go in the future.

 

TLDR of my take:  AI uses purely historic weather observations, whereas traditional NWP uses purely Newton's Laws of physics, with approximations. 

I'm sure someone at the MetO is playing with AI but they surely aren't using it yet.

 

Really helpful post. One to remember for the future, and answers questions as well as confirming observational conclusions. Friday isn’t going to be a simple thing to get correct.

[Purga]

What's sad for me is how the formidable block of very deep cold air see below

gets so easily annihilated in a couple of days to this pathetic specimen

with all the talk about 'how notoriously hard cold blocks are to move' etc, etc...

It seems like this one will be a real powder puff of pushover!

It's not happened yet of course and there's still a lot to be resolved but it still looks like the set-up which promised so much has feet of clay.

Hope it goes as well as it can...

[shotski]

48 minutes ago, Nick_Somerset said:

Ian Ferguson local BBC forecast using MO model showed good snow event for Thur/Fri but with the last words "we (SWest) may miss it... but I wouldn't guarantee it".. mmmm the plot thickens.

Interesting temperature anomalies for Friday, 850’s above but surface temperatures still way below. It’s going to be interesting. 

[johnny1972]

2 minutes ago, Purga said:

What's sad for me is how the formidable block of very deep cold air see below

gets so easily annihilated in a couple of days to this pathetic specimen

with all the talk about 'how notoriously hard cold blocks are to move' etc, etc...

It seems like this one will be a real powder puff of pushover!

It's not happened yet of course and there's still a lot to be resolved but it still looks like the set-up which promised so much has feet of clay.

Hope it goes as well as it can...

Couldn’t agree more ??

[stodge]

Evening all

To respond to the comments about the cold pool, have a look at these from the GFS 12Z Control run:

This is where we are with a long pool of cold air stretching right back to Siberia and beyond.

Now, to T+42:

Note how the chain is snapped to the east of us - the HP over Scandinavia now splits - one part retrogresses to Greenland, the other moves away ESE. There's nothing to maintain the quality of the cold pool until later in the run.

[AWD]

I'm tired, so I may be being a bit thick here, but I'm confused a little.

These are the 12z EC ens spread for the Dorset/Somerset area.  The 12 EC Det is the black line.

As I understand it, the 12z EC Det had positive 850hpa temps for a short while around the early hours of Friday morning comfortably over Dorset.  Yet, the black line, representing the EC Det run clearly does not breach the 0c Isotherm line throughout, including Friday morning.  Am I correct or am I missing something here?

Edited February 26, 2018 by AWD

[Summer Sun]

12z ECM ens for London has brought the change to less cold air forward 24hrs it was the 1st into 2nd it is now the 28th into 1st

12z

00z

Edited February 26, 2018 by Summer Sun

[Alexis]

13 minutes ago, Purga said:

What's sad for me is how the formidable block of very deep cold air see below

gets so easily annihilated in a couple of days to this pathetic specimen

with all the talk about 'how notoriously hard cold blocks are to move' etc, etc...

It seems like this one will be a real powder puff of pushover!

It's not happened yet of course and there's still a lot to be resolved but it still looks like the set-up which promised so much has feet of clay.

Hope it goes as well as it can...

The visuals don’t help. Only -10 850s on Wetterzentrale are blue, which just makes 0 to -8 temperatures an indistinction of green. Change the -8 and -9 temperatures to blue and the cold block will reappear.

[Stuie]

4 minutes ago, Summer Sun said:

12z ECM ens for London has brought the change to less cold air forward 24hrs it was the 1st into 2nd it is now the 28th into 1st

12z

00z

Interesting charts, obviously the same for my neck of the woods apart from peaks at about -2/-3. That makes no difference to the end of the week though (Friday) as it is the same outcome.

[bluearmy]

Eps have a spread which more reflects the icon days 6/7. 

Wonder if it will show as a discreet cluster later on ? 

[Stuie]

BA, can you point me towards your clusters please mate?

[pyrotech]

those wondering about tonight

here are current and predicted radar and Cardiff looks goos around 11:30

http://www.floodwarn.co.uk/live snow reports snowing now.htm

[Summer Sun]

Northwest Iceland hit a balmy 8c today looks like normal service will resume for them next week as bitter air returns along with lower pressure eventually

For us, any flow down from the north or northwest will have an edge to it albeit not to the extent of this week's cold snap

 

Edited February 26, 2018 by Summer Sun

[stratty]

1 minute ago, pyrotech said:

those wondering about tonight

here are current and predicted radar and Cardiff looks goos around 11:30

http://www.floodwarn.co.uk/live snow reports snowing now.htm

Those predicted charts are if the precipitation stays constant at that time you looked at it. Unfortunately the showers are dying out currently around Northants! 

[Purga]

7 minutes ago, bluearmy said:

Eps have a spread which more reflects the icon days 6/7. 

Wonder if it will show as a discreet cluster later on ? 

To paraphrase John Cleese --- ' It's not the despair, Nick. I can take the despair. It's the hope I can't stand!'

 

 

[chionomaniac]

36 minutes ago, Purga said:

What's sad for me is how the formidable block of very deep cold air see below

gets so easily annihilated in a couple of days to this pathetic specimen

with all the talk about 'how notoriously hard cold blocks are to move' etc, etc...

It seems like this one will be a real powder puff of pushover!

It's not happened yet of course and there's still a lot to be resolved but it still looks like the set-up which promised so much has feet of clay.

Hope it goes as well as it can...

I said that the other day - it is not the cold air but the block that we should concentrate on - when that vanishes the cold air goes with it. Wouldn't write anything off yet though because the blocking pattern may be mis modeled.

[bluearmy]

11 minutes ago, Stuie W said:

BA, can you point me towards your clusters please mate?

 

Edited February 26, 2018 by bluearmy