Cold hunt - models and chat

[snowangel32]

6 minutes ago, Daniel Smith said:

Well..

Goodluck to anybody attempting to make a forecast for later next week. This is getting a little ridiculous now 

Just a tad scattered mate. But shows nothing is resolved. FI is probs 48 hours given the positioning of that low.

Edited January 20, 2019 by snowangel32

[Weathizard]

Yesterday we had the Giraffe, seahorse and the dog... now we have a rabbit, very strange vortex profiles being modelled

[Roger J Smith]

OUTLOOK: Snow or rain in cold or mild east, north or west winds, temperatures between record lows and record highs, could clear for sunshine and frosts, otherwise very mild nights with fog. Low UV index. 

[Met4Cast]

Yesterdays cluster show the Easterly solution now in the minority with more support for the flatter pattern, though it's hardly a real majority with many possibilities still on the table.

A slight shift towards a colder pattern from this mornings ICON, ECM too to some extent before it sends a little too much energy East. 

I think we can forget about that Easterly now it's probably not going to happen, onto looking for the next source of cold. Will be interesting to see where the Met stand in todays update

Edited January 20, 2019 by Daniel Smith

[Mizzle]

4 minutes ago, Weathizard said:

Yesterday we had the Giraffe, seahorse and the dog... now we have a rabbit, very strange vortex profiles being modelled

Actually. It's a hare. Model fatigue confusion no doubt.

[Jackski4]

1 minute ago, Mizzle said:

Actually. It's a hare. Model fatigue confusion no doubt.

The blue looks like some kinda chicken to me..

[Spikey M]

12 minutes ago, Weathizard said:

Yesterday we had the Giraffe, seahorse and the dog... now we have a rabbit, very strange vortex profiles being modelled

I'm pretty sure it's a man in the recovery position with a knife sticking out of his forehead. Which, if these models don't sort themselves out, may be telling my future.

[Winterfun]

Have you not yet read the mo update yet then ?

[snowfish1]

29 minutes ago, Jackski4 said:

When ICON went against the easterly a day or so ago, it took 2 ecm runs to catch up and everyone was then disappointed. Hopefully with ICON looking better again, the ECM will catch up later on today!

I LIKE your thinking!! 

[I remember Atlantic 252]

15 minutes ago, Weathizard said:

Yesterday we had the Giraffe, seahorse and the dog... now we have a rabbit, very strange vortex profiles being modelled

and the d*ck

[SLEETY]

Another week going to go by without any meaningful snow and still no solid northern blocking. Think the ec46 and other models  forecasting HLB throughout 2nd half of  Jan and Feb which has been showing since early December are looking like going bust in their forecasts

 

[snowfish1]

8 minutes ago, Jackski4 said:

The blue looks like some kinda chicken to me..

With big. Puffed out cheeks 

[Jackski4]

1 minute ago, snowfish1 said:

With big. Puffed out cheeks 

Right!! I knew I wasn't the only one who saw that... I might be just wishful thinking about the ECM later but I do have a gut feeling!

 

Mind you, we are all sat here finding animal shapes in weather model charts, so I'm not too sure any of us are mentally stable enough to call it as it stands!

[IanT]

38 minutes ago, SLEETY said:

 

 

So close you can't rule it out with the back tracking going on this morning. Terrible performance by all the models whatever the final outcome. 

Terrible performance? 

What is going to happen next week may well have changed. To suggest that a forecast updated by 48h more data is “terrible” is a viewpoint underpinned by an implicit assumption that the weather is a deterministic system in which no randomness is involved in the development of future states. A deterministic system being one that will always produce the same outcome from a given starting condition or initial state.

But I think that we all understand innately that weather is far from deterministic. There are chaotic elements - the proverbial light left on in Fairbanks etc. - which influence outcomes. As forecast horizons extend the scope for these to have an impact increases. Given a specific set of starting conditions, multiple outcomes are possible. The weather that we get is one realisation, but there may have been other equally plausible outcomes at the point the forecast was made.

Criticism of NWP outputs on this board (and others!) often follows a pattern. Posters will say that the models are "struggling to tie down" a feature, or that the models are "always poor in this type of setup". With the benefit of 20/20 hindsight, posters will say the model was "wrong because it failed to pick up ... [height rises over the Azores?]".  Any verification gap is thus attributed to modelling weakness/failure, overlooking the possibility that something changed between the point at which the forecast was made and its realisation.

What is the point of all of this? I guess it's a plea for acceptance that the models aren't "searching for a solution". We shouldn't criticise them when they "flip-flop" from one outcome to another, especially where longer time periods are involved. There simply isn't one path that the models have to "lock-on" to. Weather has chaotic elements. If the models this afternoon don't show the extended cold spell they were showing yesterday it might not be because they are inaccurate. Something might well have changed!

NWP models are wonderful tools for forecasters. I'm sure that they have massively improved forecasting accuracy in recent years, and that these improvements will continue. But we should always remember that weather is not a deterministic system, and modulate our expectations and responses accordingly.

[KTtom]

Check the tweets thread for why the met haven't change their forecast...Yet.. 

Yes, were back to day 10 again but signs we may finally get an Atlantic block ?

 

 

[Jackski4]

To my knowledge, about an hour until we get a fresh ICON run followed by the GFS 6z at 9:30.. 

 

As I said, I prayed to the weather gods, hopefully they've listened.

[SLEETY]

9 minutes ago, IanT said:

Terrible performance? 

What is going to happen next week may well have changed. To suggest that a forecast updated by 48h more data is “terrible” is a viewpoint underpinned by an implicit assumption that the weather is a deterministic system in which no randomness is involved in the development of future states. A deterministic system being one that will always produce the same outcome from a given starting condition or initial state.

 

Wow long reply or what. I cut the rest out The simple question I and many would like to know is why if the weather is Zonal Atlantic driven the models can quite accurately forecast several days ahead without too much trouble, but if the weather attempts to come in from the other direction, it causes no end of confusion in the models. Look at the past few days for example. 

Why should the direction the weather comes to us affect the accuracy of the weather models so dramatically. 

 

[Met4Cast]

5 minutes ago, SLEETY said:

Wow long reply or what. I cut the rest out The simple question I and many would like to know is why if the weather is Zonal Atlantic driven the models can quite accurately forecast several days ahead without too much trouble, but if the weather attempts to come in from the other direction, it causes no end of confusion in the models. Look at the past few days for example. 

Why should the direction the weather comes to us affect the accuracy of the weather models so dramatically. 

 

It's all to do with climatology. The default weather pattern for our part of the world is Westerly winds and low pressure systems coming in on the jet stream, models handle this well because it's the "norm" and tends to be quite predictable. Low pressure moving West, work out it's speed and where the future forcings on the system will be and it's fairly easy to "guess" accurately where that system might be in 1,2,3,4 days time. 

Now take that same system but instead of sending it Westwards like normal, stick a block of high pressure in front of it, not only that but also split the energy from the low into different directions. Now the model needs to handle not only the track of the system, but the systems speed varying, the energy from the system going in 2 directions as it splits, the strength of the high, where will wind direction be at any given moment across the UK? 

Models can generally handle this quite well when looking at the NH pattern as a hole, but given the UK is a tiny island in one corner of the hemisphere it doesn't take a huge change in direction/speed/split energy to have a big effect on the weather we see here on the ground.

Now throw into the mix the MJO, downwelling from the SSW, El Nino etc and it gets infinitely more complicated. Plus the models probably do struggle with zonal at times but how many of us can honestly say we micro analyse the models when zonal is on the cards the same way we do when cold and snow are on the cards?

Edited January 20, 2019 by Daniel Smith

[Radiating Dendrite]

5 minutes ago, SLEETY said:

Wow long reply or what. I cut the rest out The simple question I and many would like to know is why if the weather is Zonal Atlantic driven the models can quite accurately forecast several days ahead without too much trouble, but if the weather attempts to come in from the other direction, it causes no end of confusion in the models. Look at the past few days for example. 

Why should the direction the weather comes to us affect the accuracy of the weather models so dramatically. 

 

1. We do not analyse zonal to the same extent, so changes are not picked up.

2. Larger margin for error with zonal.

Edited January 20, 2019 by Radiating Dendrite

[Firefly2005]

48 minutes ago, IanT said:

Terrible performance? 

What is going to happen next week may well have changed. To suggest that a forecast updated by 48h more data is “terrible” is a viewpoint underpinned by an implicit assumption that the weather is a deterministic system in which no randomness is involved in the development of future states. A deterministic system being one that will always produce the same outcome from a given starting condition or initial state.

But I think that we all understand innately that weather is far from deterministic. There are chaotic elements - the proverbial light left on in Fairbanks etc. - which influence outcomes. As forecast horizons extend the scope for these to have an impact increases. Given a specific set of starting conditions, multiple outcomes are possible. The weather that we get is one realisation, but there may have been other equally plausible outcomes at the point the forecast was made.

Criticism of NWP outputs on this board (and others!) often follows a pattern. Posters will say that the models are "struggling to tie down" a feature, or that the models are "always poor in this type of setup". With the benefit of 20/20 hindsight, posters will say the model was "wrong because it failed to pick up ... [height rises over the Azores?]".  Any verification gap is thus attributed to modelling weakness/failure, overlooking the possibility that something changed between the point at which the forecast was made and its realisation.

What is the point of all of this? I guess it's a plea for acceptance that the models aren't "searching for a solution". We shouldn't criticise them when they "flip-flop" from one outcome to another, especially where longer time periods are involved. There simply isn't one path that the models have to "lock-on" to. Weather has chaotic elements. If the models this afternoon don't show the extended cold spell they were showing yesterday it might not be because they are inaccurate. Something might well have changed!

NWP models are wonderful tools for forecasters. I'm sure that they have massively improved forecasting accuracy in recent years, and that these improvements will continue. But we should always remember that weather is not a deterministic system, and modulate our expectations and responses accordingly.

Really interesting post! Quite philosophical... thanks

[MattStoke]

56 minutes ago, SLEETY said:

Another week going to go by without any meaningful snow and still no solid northern blocking. Think the ec46 and other models  forecasting HLB throughout 2nd half of  Jan and Feb which has been showing since early December are looking like going bust in their forecasts

 

The ECM 46 day model isn’t showing HLB until the week of the 28th of January.

[Met4Cast]

58 minutes ago, SLEETY said:

Another week going to go by without any meaningful snow and still no solid northern blocking. Think the ec46 and other models  forecasting HLB throughout 2nd half of  Jan and Feb which has been showing since early December are looking like going bust in their forecasts

 

Depends..

EC46 for the 22nd vs GFS forecast for the 22nd

Broadly the lows/highs are where the EC46 predicted. EC46 doesn't really show any meaningful HLB until the 1st week fo February

 

 

[smith25]

4 minutes ago, MattStoke said:

The ECM 46 day model isn’t showing HLB until the week of the 28th of January.

To be fair i thought the main chances of wintery weather was going to start early feb and anything before then was a bonus.

[Met4Cast]

I'll post this chart to hopefully calm some nerves

EPS continuing to highlight below average temperatures right out into the extended - Despite a "less cold" blip possibly later next week now looking likely the overall theme and consensus across the models is for temperatures to remain below average with a trough across Europe. The first Easterly may have failed (may not have yet, the spread is so huge there's a possibility it'll come back, ICON seems to think so) but there is already another sinking trough followed by Atlantic height rises showing on the models for days 7-11. 

[bluearmy]

55 minutes ago, SLEETY said:

Another week going to go by without any meaningful snow and still no solid northern blocking. Think the ec46 and other models  forecasting HLB throughout 2nd half of  Jan and Feb which has been showing since early December are looking like going bust in their forecasts

 

The ec 46 charts WE see for free on here are 7 day means ..... they won’t pick up the two day blip of transient upper ridge passing through next weekend .....the average onset of high heights to our nw and low heights to our se combo was around 24 jan as the mid point 

looking at this chart (it’s a five day mean centred 23/1), I’d say it did a bloody good job and the sneak through of the low heights through the centre of the upper ridge just isn’t something you can accout for in LR modelling ..... it’s called weather and it means the ridging to the north loses its feed from that to the south .... you have seen what happens because of that over the past 48 hours with the collapse of the pattern we were expecting once the trough dropped through 

 

 

If if anyone doesn’t understand that this proves the ec46 to have be pen a very good tool for this part of winter then they need to think twice before posting on more technical stuff ....