Model Output Discussion 18/11/2012 12Z onwards

[chionomaniac]

History tells me that when the gfs shows something like this in FI it can be an indicator that change is backtracking just a little.

Of course the mean and trend is what is important but the operational run cannot just be swept aside.

As I have just said JS, the vortex is not going to stay split forever - so of course any model that goes out to 16 days is going to try and find a solution of how it is likely to reform!

[nick sussex]

It always makes me laugh when people start worrying over a possible breakdown in the expected blocking pattern before its even arrived on the scene.

Lets just get there and see how the models evolve.

In terms of todays output good agreement in terms of the transition to higher pressure to the north and nw, I particularly like the ECM as its got more margin for error than its previous operational runs.

[Bristawl Si]

History tells me that when the gfs shows something like this in FI it can be an indicator that change is backtracking just a little.

Of course the mean and trend is what is important but the operational run cannot just be swept aside.

History tells me that when the gfs shows something like this in FI it can be an indicator that change is around the corner but it's just trying to work out exactly what comes next.

If Stewart, who IS one of the 'true' experts on here says follow the mean and trends as the Ops past a certain timeframe can mean "diddly squat" then that's good enough IMO.

[london-snow]

Great post nick my thoughts very much the same.

As i remember 2010 - 2011 the famous event itself. A few op runs were showing breakdowns even before we got the cold air in place however once we had the cold air established the breakdown was put further and further back.

I feel that if we get over the first hurdle we then have the winning post of the grand national in sight.

[AWD]

The ensembles are trending colder & colder IMO. Bear in mind these are for London, there are a large amount heading down to -5c now.

http://www.meteociel.fr/cartes_obs/gens_display.php?ext=1&x=306&y=141&ville=Londres

Edited November 20, 2012 by AWD

[latitude]

Maybe its because the models are not showing any -5 850's until FI.

[Robbie Garrett]

Just got a response from the MetOffice. Hopefully this answers a few questions... from a proffesional point of view.

Thank you for your email.

I think you are right in that we are being cautious. I think we all can see it is getting colder towards the end of next week, but it is a question of how much colder.

The evidence at present does not indicate anything particularly severe, cold yes, but nothing we wouldn't normally expect from a cold snap late Nov/early Dec.

When you have headlines in certain daily newspapers indicating the coldest winter for 100 years, we need to make sure we give out the right message between a typical early winter cold snap and possibly something more substantial and that is where the uncertainty lies.

We will be firming up on these details over the next few days and hopefully we will become more confident.

I hope you find my comments of use.

Edited November 20, 2012 by Robbie Garrett

[nick sussex]

Great post nick my thoughts very much the same.

As i remember 2010 - 2011 the famous event itself. A few op runs were showing breakdowns even before we got the cold air in place however once we had the cold air established the breakdown was put further and further back.

I feel that if we get over the first hurdle we then have the winning post of the grand national in sight.

Thanks. In terms of model bias you will often see the GFS trying to remove blocking in the lower resolution output. What often happens is that this never makes it into the higher resolution.

Blocking patterns have to evolve, you cannot for example have a Greenland block in the same position for weeks, equally Scandi highs also evolve.

Generally when the atmosphere is predisposed towards higher latitude blocking you often get a retrogression signal, for example if people look at the archive chart for end of November 2010 and follow them through you'll see the shuffling from Scandi high to , Greenland block,then you'd get a briefish easterly again as the ridge gets flattened but then back to Greenland yet again.

For Europe in these circumstances if you avoid a western based negative NAO during that process then you generally stay cold.

I think people sometimes expect too much from numerical weather prediction, the models are often slow to pick up on signals or when they do they can often over react.

Operational output is at its worst when a pattern change is imminent, its best to wait for the models to settle down once the block is in place and at that point we can see how the pattern could develop.

Edited November 20, 2012 by nick sussex

[VillagePlank]

Before anyone gets too excited, here's a subsequent analysis of the GEFS output, this morning,

This is for every six hour period since midnight last night until midnight 5th Dec. The blueline is the coldest run of the ensembles, the redline is the warmest line of the ensembles. Green is the mean, and the space between the dotted black lines is the 68% probability area - ie 68% of the time the temperature will fall between these two lines.

We just about skirt the -5C 850hPa way way off, and it looks like the coldest run is skewing the chart to show more cold potential than there really is.

(Corrected for wrong percentages)

Edited November 20, 2012 by Boar Wrinklestorm

[latitude]

So from a newbie perspective we can use FI to see the trends for a potential cold spell (which we have over the last few days or more) but we cannot use FI to see the trends of a milder pattern evolving.

Why cant this cold spell just be a short affair and the GFS now showing and trending towards mild soon after?

[AWD]

In the short term there looks like there could be significant rainfall totals this week in some western locations;

http://expert-images.weatheronline.co.uk/daten/proficharts/en/2012/11/20/basis00/ukuk/rsum/12112600_2000.gif

Further flooding issues likely I fear in the next few days.

[Polar Maritime]

So from a newbie perspective we can use FI to see the trends for a potential cold spell (which we have over the last few days or more) but we cannot use FI to see the trends of a milder pattern evolving.

Why cant this cold spell just be a short affair and the GFS now showing and trending towards mild soon after?

It could just be a short lived affair, but at the moment it's not. And there is nothing in the charts trending towards mild soon after, the odd run is not a trend.

Let's wait for the cold pattern to arrive first, and then we shall see.

Edited November 20, 2012 by Polar Maritime

[Thundery wintry showers]

The ECMWF ensemble mean has been pretty consistent over the last couple of days in showing a north to north-easterly flow over the British Isles and 850hPa temperatures close to -5C (bearing in mind that the ECMWF often overestimates 850hPa temperatures, whereas GFS often underestimates them). From this, there would be no widespread snow cover at low levels but many places, especially in the eastern half of the country, may see sleet and snow showers. I remember sleety showers here at Sandhutton on 27th October with 850hPa temperatures having risen to around the -4C mark.

Before this, as well as the issue of high rainfall totals in the west, Thursday and into early Friday look like seeing a zone of strong to gale force winds extend eastwards across the British Isles- although I think GFS may be over-egging the depression a little as the UKMO/ECMWF have the isobars slightly further apart.

I see little evidence, from the model outputs or the teleconnections, that we won't keep a blocked pattern well into December, but as always it is a question of where the blocking ends up re. if we are to get widespread snow cover.

Edited November 20, 2012 by Thundery wintry showers

[Suburban Streamer]

could someone give me the link to the nao/ao spaghetti forecasts.

the link i usually use doesnt seem to be updating...

http://policlimate.c...scillation.html

thanks

Edited November 20, 2012 by Suburban Streamer

[pyrotech]

Fistly what a well balanced and honest reply from meto.

Secondly lets not forget the hype off the express about 10 days ago.

In my opinion the gfs did first indicate a change in trend. It now has dropped it for original timeframe but trend still remains in fi.

Do not expect any real similarity between what gfs originally showed in actual weather but do expect something resembling the trend it howed. You notice trend for much cooler remains in fi on gfs.

What often happens now is timeframe for ecm comes into play for same time gfs thought change would occur, Ecm picks it up and very late Gfs brings it back out of the blue in reliable frame. The trend by gfs is normally a good indicator deep in fi but rather poor for some unknown reason inbetween. Its not always like this with gfs finding correct change in trend but occasionally its very early and gets something fairly similar too the change. Its often a little too early by a day or two and often gets location of cold out by few hundred miles. But the change is indicated which is then left for us to search for posabilities from the change. Although often gfs fi can be written off as useless, i do think the basics for change it shows can be useful as a tool. Forecasting for snow in gfs fi is foolhardy but forecasting blocking and possible cold plumes in gfs fi is too a certain extent possible. Even with the blocking in place we can still be stuck in mild air. That does not mean the blocking trend first shown was wrong. Just that orientation was not right for us. I expect now ecm to start showing something like gfs showed a week ago and then gfs to jump back on line. Also positive that gfs continues cold in fi as i see reload potential. Liam. Polar kold ,chino ,steve and jh amongst others have been here a long time giving constructive and unbiased posts new users would do well listening to their thoughts

[latitude]

Excuse me?? Trying to learn how things work here. MODS that is out of line.

LOL! You just keep on failing there failtroll

Thank you for the reply.

It could just be a short lived affair, but at the moment it's not. And there is nothing in the charts trending towards mild soon after, the odd run is not a trend.

Let's wait for the cold pattern to arrive first, and then we shall see.

[phil nw.]

The signs are recent GFS Op.runs have been out of kilter with mean outputs as we enter into lower resolution.That is the period where the pattern change will further develop but only the GFS later frames go out that far and that`s why the ECM will naturally show less volatility only going as far as 10 days.

To illustrate how the GFS Op. can go awol later on here are some graphs that show the Ens.outputs are still trending negative for the NAO/AO states in the medium term onwards.

we can see the Operational(green) index line goes positive much against the rest of the suite-particularly the AO index..

The GFS/ECM t240hrs ht. anomls. show the change of pattern establishing.

Building of cross polar hts showing up nicely.

I would be more concerned if the Operational outputs showed the blocking weakening within it`s earlier frames but these look pretty decent to me and compares well here the GFS/ECM Ops. at the same time.

Later frames of the GFS will likely continue to show different solutions in the placing the segments of a vortex split or displacement for now.Perhaps it`s best to compare runs against the ens mean outputs and trendings untill closer to real time.

It would be better than feeling the need for some of Nick`s prozack.

[VillagePlank]

According to the chaps over at CPC, the NAO has a near infinite amount of possibility as we head into early December,

Edited November 20, 2012 by Boar Wrinklestorm

[Tim Bland]

According to the chaps over at CPC, the NAO has a near infinite amount of possibility as we head into early December,

All bar one go negative though which is all that matters IMO

[VillagePlank]

Yes, of course.

The curious thing is that if one gets the monthly historic 500hPa NAO values, and the CET monthly values they have a correlation of 0.004 (between -1 and 1) which tends to the idea that these two variables have no linear dependence between them. There is a slight signal that if the NAO increases in value, then so does the CET, but absolutely no confidence in it.

Drawing a chart demonstrates this idea,

I realise that is nominally said that a negative NAO tends to a low CET, for instance Winter 2009-2010 coincided with a negative NAO, but Octobers NAO was even more negative and the CET was above average.

I suspect, as always, that it is in combination with other factors where it's role becomes important.

Edited November 20, 2012 by Boar Wrinklestorm

[VillagePlank]

A stronger case can be made for the Arctic Oscillation (AO),

Not that much stronger, statistically, but an order of magnitude stronger than the NAO. The AO also looks like it affects us once it is in the extremes - ie it is a better indicator of cold when it is below -2, and better indicator of warmth when it is above +2. In between those values there really isn't any indicators at all.

And the good news ....

Down she goes .... based on this, and this alone, I'd forecast the CET to plummet around the turn of the month.

Edited November 20, 2012 by Boar Wrinklestorm

[summer blizzard]

I think that some people need to take a more pragmatic view of GFS FI and remember 2 things..

1) It's FI

2) Whilst it may be poor for snow, not once does the upper air breach 0C for the bulk of the UK and with high pressure we will likely see strong inversions.

In terms of the modelling the ECWMF has significant potential with a cold pool to the far east slowly moving westward and the trough to our south elongating in such a way as to drag a more northerly element down with subsequent uppers.

[weather eater]

So from a newbie perspective we can use FI to see the trends for a potential cold spell (which we have over the last few days or more) but we cannot use FI to see the trends of a milder pattern evolving.

Why cant this cold spell just be a short affair and the GFS now showing and trending towards mild soon after?

You can do either if you want, but the trend in a particular model is not the be all and end all of making a call for a particular scenario to develop. Trends on the models come and go all the time, sometimes the models will run with an idea for a while and then drop it, and this seems especially true of the GFS, perhaps because it goes out so far. However the experts when calling a trend are looking beyond a few runs on the GFS, especially as more tools are becoming available to them, a few runs on the GFS won’t bother them until such time as support comes from the other diagnostic tools. The move to a blocked spell is currently well supported if that then changes then so will the view of the more knowledgeable members. I would advise newbie’s and the less knowledgeable that they should only pay attention to those who seem to know what they are talking about and who back it up with evidence. Because that’s what I do, then apply my own logical and pragmatic thinking, I really can’t be bothered with all this jumping up and down getting my knickers in a twist over models that I know to be extremely fallible beyond 144hrs max.

Edited November 20, 2012 by weather eater

[Yozzer]

I realise that is nominally said that a negative NAO tends to a low CET, for instance Winter 2009-2010 coincided with a negative NAO, but Octobers NAO was even more negative and the CET was above average.

I suspect, as always, that it is in combination with other factors where it's role becomes important.

Surely one of the major factors in trying to corelate NAO with CET will be the time of year and the source of the airmass. In the above example any continental flow with a -NAO would give a higher than average CET in October when the continent is still warm, but conversely a lower one come the winter quarter, or with an arctic airmass. Or am I being simplistic here???

[VillagePlank]

I don't think you are being too simplistic. A great number of things affect our weather here in Britain, and it would come as no great surprise if any correlation came with conditional parameters. I think it is overly simplistic, however, to look at the NAO signal, and if negative, predict cold, and, to a lesser degree do the same with the AO. It already seems that the AO is utterly useless if it's between -2 and 2; and contrary to my initial thoughts AO outside those boundaries tend to predict coolness with a negative value outside of the boundaries a predictor of coldness

More in a bit.

Edited November 20, 2012 by Boar Wrinklestorm