There seems to be a problem that the more detailed a model becomes, the more accurate it is in the near term, but the likelier it is to go haywire in the longer term. We're all familiar with those  phantom day 10 ECM easterlies over the last couple of years, and the new GFS, in its attempt to be as detailed, may have gone the same way.

My guess (and it's purely an uninformed guess) is that Numerical Weather Predictions, which is what these models are, will always face a problem with making the grids smaller and the time-steps more numerous. Out to day 5, say, this is a good thing. But beyond that we're talking about an accumulation of tiny errors which become huge errors as they're carried forward from one calculation to the next. Even something as innocuous as a rounding 'error', such as calculating air pressure as 1020.5 mb and carrying that number forward to the next 3-hour interval, might produce northern blocking by day 10 if the real figure is 1020.54 bar, or zonality if the real figure is 1020.45 bar.

And the more of these tiny rounding errors you make as you increase resolution, the greater the scope for critical errors to accumulate later on.

I'd be interested to hear from anyone who does know about these things whether I'm barking up the wrong tree or not here. 

I'm no expert, but I don't see why improving a model for the short term should make it less accurate in the long time. There might well come a point at maybe around 10 days out when the accumulation of random errors means it is no longer any better than it was before it was improved, but I don't see why it should actually be worse.

---

Says alot about our winter when i look at a disntinctly zonal outlook (beyond this week) and see that as an upgrade to cold weather.

At least the mountains will see something seasonal.

---

Not looking great after the one to two day colder blips. Still we will increase our blip count to three for this winter. This could be my third consecutive winter with no snow lying.

---

There seems to be a problem that the more detailed a model becomes, the more accurate it is in the near term, but the likelier it is to go haywire in the longer term. We're all familiar with those  phantom day 10 ECM easterlies over the last couple of years, and the new GFS, in its attempt to be as detailed, may have gone the same way.

My guess (and it's purely an uninformed guess) is that Numerical Weather Predictions, which is what these models are, will always face a problem with making the grids smaller and the time-steps more numerous. Out to day 5, say, this is a good thing. But beyond that we're talking about an accumulation of tiny errors which become huge errors as they're carried forward from one calculation to the next. Even something as innocuous as a rounding 'error', such as calculating air pressure as 1020.5 mb and carrying that number forward to the next 3-hour interval, might produce northern blocking by day 10 if the real figure is 1020.54 bar, or zonality if the real figure is 1020.45 bar.

And the more of these tiny rounding errors you make as you increase resolution, the greater the scope for critical errors to accumulate later on.

I'd be interested to hear from anyone who does know about these things whether I'm barking up the wrong tree or not here. 

I think there are two main problems.  

As I've observed before, one is that the greater level of detail in the newest models means there are more datapoints to populate for the starting position and there simply aren't accurate readings for every one of them. This means the opening position still contains estimates/approximations.  They may be very good estimates but they're still estimates. That is why the ensemble suites are still critically important.

The other issue is that however good the models are, they are still only a mathematical representation of how a complex fluid sitting on a complex surface will behave. 

I don't know how many decimal places are used in the models but I'd be surprised if that was an issue and certainly not significant relative to these other two.

I don't see any logic for the programming giving better short-term results but worse longer term ones. That's not how the models are constructed: the same mathematical formulae are used throughout.

---

Not really that bothered but these ensemble plots are poor considering it’s mid winter. OK for elevation at times but pants for lowland UK.
Whether the rebounding COD will save the day or a warming remains to be seen but January is pretty much a bust again this year.

---

Not really that bothered but these ensemble plots are poor considering it’s mid winter. OK for elevation at times but pants for lowland UK.
Whether the rebounding COD will save the day or a warming remains to be seen but January is pretty much a bust again this year.

http://old.wetterzentrale.de/pics/MT8_London_ens.png

(probably not worth saying twice but a poor signal doesn’t help :))

 

 

---

I actually quite like the look of the 06z GFS. No deep cold but chilly and lively, which is far preferable to dull and dry imo. 

Having said that today is beautiful. I love a clear high pressure but after today lots of cloud gets pulled in and it looks dull and mild. Incredibly boring weather unless you have to work outside.

I actually quite like the look of the 06z GFS. No deep cold but chilly and lively, which is far preferable to dull and dry imo. 

Having said that today is beautiful. I love a clear high pressure but after today lots of cloud gets pulled in and it looks dull and mild. Incredibly boring weather unless you have to work outside.

 

It's already been pulled in here. I actually like dull weather when it's cold, but dull and mild is just 

Anyway, the Manchester ENS show the end-of-Jan colder blip to be a couple of degrees cooler than London on the 850's (a mean of -6c as opposed to -4c).

Interesting to note that for both Manchester and London there is strong agreement between all members for the dip on the 29th/30th (on both, the difference between mildest & coldest is only 2c), yet the earlier and smaller dip on the 26th is far less certain, especially for London, which has a scatter of about 5c

---

The promise of a blink and you’ll miss it colder blip at the end of the month which will more than likely flatten out anyway doesn’t really rock my boat. This decrepit winter is a bust, I’m looking forward to plume watch now and the first signs of spring warmth. Having said that the 06z GFS should deliver some snow to Scottish mountains for the ski industry, for how long it would hang around however is uncertain

The 12z GFS is actually better than the 06z. In the context of this Winter I'd take it with open arms. Plenty of PM shots and snow for northern areas with elevation and perhaps down to lover levels at times too. 

The ECM ens were horrific earlier so no doubt it will be along to burst this little bubble of hope shortly.

The 12z GFS is actually better than the 06z. In the context of this Winter I'd take it with open arms. Plenty of PM shots and snow for northern areas with elevation and perhaps down to lover levels at times too. 

The ECM ens were horrific earlier so no doubt it will be along to burst this little bubble of hope shortly.

Whereas in my eyes,  next week is looking awful with widespread strong winds on Monday and a storm system slicing through the UK on Thursday. That holds no appeal to me at all but given that the end of next week is a long way off in forecasting terms and thus is subject to change anyway - which hopefully include downgrades.

---

Are we in for a back loaded Winter I wonder ?? 

---

Are we in for a back loaded Winter I wonder ?? 

If you mean less stupidly mild than the first half then probably. The jury is still out on whether it includes anything wintry however.

---

Are we in for a back loaded Winter I wonder ?? 

Back loaded into April probably!

---

https://www.wetterzentrale.de/en/topkarten.php?map=1&model=ecm&var=1&run=12&time=192&lid=OP&h=0&mv=0&tr=24#mapref

If modelled correctly, the HP situated East of Canada and the US could be interesting..

Long way out though. 

ECM 192

Not looking great after the one to two day colder blips. Still we will increase our blip count to three for this winter. This could be my third consecutive winter with no snow lying.

Lying? So far there is no sign of anything even falling away from high ground. Although you must have been incredibly unlucky not to get any lying snow last year or 2018 (or are you only classing meteorological winter and discounting March 2018?)

---

 

Lying? So far there is no sign of anything even falling away from high ground. Although you must have been incredibly unlucky not to get any lying snow last year or 2018 (or are you only classing meteorological winter and discounting March 2018?)

I don’t really remember any lying snow last year here. I hardly remember any falling snow for that matter.

---

Are we in for a back loaded Winter I wonder ?? 

If Judah is correct, possible methinks  

“There hasn't been much opportunity to tweet about the #polarvortex (PV) the past month but finally some signs of change. The biggest vertical energy pulse (needed to disrupt the PV) in over a month predicted resulting in a stratospheric -AO for next week. More disruption possible”

Are we in for a ANY Winter I wonder ?? 

Corrected 😉

 

I'm no expert, but I don't see why improving a model for the short term should make it less accurate in the long time. There might well come a point at maybe around 10 days out when the accumulation of random errors means it is no longer any better than it was before it was improved, but I don't see why it should actually be worse.

My thinking is that the improvement in short-term accuracy comes from an increase in the number of data points, along with a narrowing of the gap between calculations as the model run progresses (say, from 12 hrs, 24 hrs, 36 hrs etc. to 03 hrs, 06 hrs, 09 hrs, 12 hrs, and so on).

The result is an exponential increase in the number of potential small errors, which (exponentially) become big errors eventually - you know, the thing about a butterfly flapping its wings and causing a hurricane.

Re Gandalf's comment that rounding errors should only have a negligible effect - I would disagree with that, particularly when dealing with our part of the world, where even the subtlest changes in a similar general pattern can result in vastly different weather on the ground. Rounding errors are the bane of iterative models (which is what these are).

A tenth of a degree centigrade here, half a millibar there, and ten days later, and thousands of calculations further on, the error gets multiplied into something substantial. But on a coarser model, with fewer data points, and a 12-hour gap between iterations, any tiny errors have less opportunity to take on a life of their own.

It's nothing to do with the mathematical calculations themselves becoming less accurate. It's just that the numbers on which those calculations are based become open to more errors as more calculations need to be made. 

---