Model Output Discussion 26/10 2012 >

[BLAST FROM THE PAST]

ECM interesting at t144, look at Finland/Lapland region it has HP ridging from east but its heights are lower over Greenland whereas UKMO has them higher....so looks in between the two to me on meteociel.

ECM then follows GFS...?

BFTP

Edited October 29, 2012 by BLAST FROM THE PAST

[BLAST FROM THE PAST]

Interesting S E BBC forecast for Friday 2nd, they see that 'zipper' to bring strong winds and very heavy rain to all of Surrey, Sussex, Kent, London

Going to keep away for a few days until say Thurs morning and see what we have in store for the weekend,

BFTP

Edited October 29, 2012 by BLAST FROM THE PAST

[AWD]

Interesting day 10 chart from the ECM

http://www.meteociel.fr/modeles/ecmwf.php?ech=240&mode=1&map=1&type=0&archive=0

Another push of heights into Greenland?

[Gibby]

Good evening folks. Here is my evening look at the 12zs from GFS, UKMO and ECM.

All models show an area of slack pressure over Britain between Low pressure areas SE and NW of the UK. The Low to the North deepens markedly over the coming days with West or SW winds freshening everywhere over the next 36 hours. most inland areas will see a lot of dry weather in the short term but frontal bands reaching the NW later tomorrow means Northern areas see rain at times tomorrow. Further South the weather looks fairly dry and bright tomorrow but as the Low to the North deepens it extends it's influence to all areas with strong winds and heavy rain through Wednesday. A cold front introduces colder and brighter weather over Thursday and Friday though a small but significant Low could move East over Southern England on Friday bringing renewed high winds and heavy rain for a time. The wekend is also shown by all models to be governed by this large deep and complex Low pressure system as it moves nowhere fast with rather cold weather with rain and showers for all continuing as well as hill snow in the North.

GFS then shows the start of next week as a steadily improving one as the Low fills and moves away East allowing a collapsing ridge to topple down over the UK on Tuesday and Wednesday. Most areas would see this period as drier and brighter than previous days with some sunshine by day and chilly misty overnight in generally rather cold temperatures. As we move into FI the weather turns milder from the West as winds back Westerly with a couple of days during the latter half of the week with a North/South split developing with occasional rain in the North and drier and mild conditions in the South. Through the latter stages of FI further cold incursions occur from the North and while pressure rebuilds over the UK in cold air some influence of a cold Northern European High could form a block from Atlantic Low pressure with a mild and windy SE flow shown in the South at the end of the run with rain threatening from the SW.

The GFS Ensembles show that a period of fairly normal 850's in Atlantic changeable conditions is the most likely outcome once we move out of the upcoming colder than average spell caused by the large Low over the UK later this week. The weather will continue changeable though countrywide throughout with occasional rain and showers. there are no indications of any trend toward notable cold or mild weather in the next few weeks anywhere in the UK.

The Jet Stream shows the flow moving SE over the UK currently before a strong surge moves across the Atlantic SE towards Northern France in the coming days. It continues South of the Uk through the rest of the week weakening briefly over the Atlantic after the weekend.

UKMO for midday on Sunday shows Low pressure over the Hebrides with a showery and unstable Westerly flow maintained over Britain with embedded troughs. Rain and showers would continue in temperatures somewhat below normal with snow on Northern hills and mountains at times.

ECM for the same time has the Low over Scotland with cyclonic winds between SW in the East and NW in the West. The cyclonic flow would ensure a continuation of rain and showers to end the weekend with snow on Northern high ground in all in all rather cold conditions. Monday then shows a trough over the South with the Low pressure finally exiting the East coast with a ridge approaching the NW highlands later bringing drier if cold conditions. As this morning's run shown the ridge crosses the UK over Tuesday with a bright and dry day after a frost overnight. By wednesday a warm front moves East over Britain bringing milder and cloudier conditions with rain and drizzle in the North while the sheltered South stays drier and mild. the end of the run shows mild and fair weather continuing in the South while Northern areas might still see some rain albeit quite mild here too.

In Summary the weather looks distinctly disturbed over the coming 7-8 days with rain and strong winds in temperatures somewhat below normal. Thereafter there are some signals for High pressure to build down to the SW of Britain pushing our deep Low pressure away East after the beginning of next week and introducing milder Atlantic Westerly winds with a North/South split the likely outcome for a while thereafter with the best of the dry weather likely in the South and East. There are also signs from GFS and ECM though that this may be just a transient phase before colder and unsettled conditions return once more from the NW.

[bluearmy]

Whilst naefs shows a flattening of the mean pattern around Greenland in the 10 day area, the ridging returns for the end of week 2, we end the run with the three anomalies remaining - Greenland high, scandi low and Aleutian/Kamchatka high. Whilst we see low anomolys near the pole week 2, the spreads indicate that the core of these likely just north of Svalbard and the fact that the 850 anomolys stay positive throughout the arctic for the whole of week 2 says to me that there is no strong vortex forming. If the meridional pattern returns after an attempt to flatten,this to me is an encouraging sign for what is likely second half of November.

[sebastiaan1973]

Gibby many thanks for yet another wonderfull written recapitulation

[snowking]

While we're talking which model is better than the other:

0z's @ Day 6:

http://www.emc.ncep....0_G2NHX_00Z.png

1. ECMWF

2. GFS

3. UKMO

4. GEM

5. NOGAPS

12z's @ Day 6:

http://www.emc.ncep....0_G2NHX_12Z.png

1. ECMWF

2. GFS

3. UKMO

4. GEM

5. JMA

6. NOGAPS

The ECMWF remains supreme at this range.

In terms of 0z vs 12z, the 0z's are correlating far better than the 12z's, perhaps suggesting that we should be paying more attention to the overnight runs. For example, both the GFS and UKMO are proving 5% more accurate on their 0z runs, and even ECMWF is at 2%.

Over a longer period of time:

http://www.emc.ncep....H500mb_day6.png

You can see that even as far back as 2000, the ECMWF has consistently remained the most accurate performer at day 6. The Ranking over the last year is as follows:

1. ECMWF

2. UKMO

3. GFS

4. GEM

5. NOGAPS

(note as this data is only available from the 0z runs, the JMA cannot be compared)

So two conclusions I personally draw from a data perspective:

1. ECMWF remains king, and has done for some time.

2. 0z runs should be taken more seriously perhaps than 12z runs for day 6 projections

3. GFS/UKMO seem fairly interchangeable for the position of runner up, and, to my mind, should perhaps be considered =2nd

SK

EDIT: Perhaps I should add significantly:

4. Modelling has continually improved through the course of the last decade, and most models have improved day 6 accuracy by between 5-10%

Edited October 29, 2012 by snowking

[summer blizzard]

While we're talking which model is better than the other:

0z's @ Day 6:

http://www.emc.ncep....0_G2NHX_00Z.png

1. ECMWF

2. GFS

3. UKMO

4. GEM

5. NOGAPS

12z's @ Day 6:

http://www.emc.ncep....0_G2NHX_12Z.png

1. ECMWF

2. GFS

3. UKMO

4. GEM

5. JMA

6. NOGAPS

The ECMWF remains supreme at this range.

In terms of 0z vs 12z, the 0z's are correlating far better than the 12z's, perhaps suggesting that we should be paying more attention to the overnight runs. For example, both the GFS and UKMO are proving 5% more accurate in their 0z runs, and even ECMWF is at 2%.

Over a longer period of time:

http://www.emc.ncep....H500mb_day6.png

You can see that even as far back as 2000, the ECMWF has consistently remained the most accurate performer at day 6. The Ranking over the last year is as follows:

1. ECMWF

2. UKMO

3. GFS

4. GEM

5. NOGAPS

(note as this data is only available from the 0z runs, the JMA cannot be compared)

So two conclusions I personally draw from a data perspective:

1. ECMWF remains king, and has done for some time.

2. 0z runs should be taken more seriously perhaps than 12z runs for day 6 projections

3. GFS/UKMO seem fairly interchangeable for the position of runner up, and, to my mind, should perhaps be considered =2nd

SK

Very surprising, i joined in 04 and right upto 07 the consensus was that GFS was superior.

[snowking]

Very surprising, i joined in 04 and right upto 07 the consensus was that GFS was superior.

Exactly the same experience I have even dating back to the old BBC Snow-watch days

I can only assume it was simply the greater range of data available that produced this belief (a belief that I too held during the years you mention)

SK

Edited October 29, 2012 by snowking

[MR EXTREMES]

well what can i say ive gone through the models gfs gefs ukmo fax and ecm and im 70% confident that something is brewing in regards to something colder.

pressure building around scandi and high pressure losing its grip around greenland but after a week also trying to rebuild although greenland heights are exciting its very possible looking east maybe a good thing.

i reckon if the vortex stays away from its home and stays as it is the i think after hurricaine sandy has weakened we may start to see more of a cross model agreement in favour of something wintry.

[mcweather]

The model verification stats at day 6 are interesting. Looking at the slightly longer term picture. There is no doubt that simply due to its longer published runs the GFS is still king at picking up the trends in what we call FI. Quite often the past four or five winters cracking lone outlier has proved actually to be a tend setter and I have often remarked on here before that even the beginings of the winters of 63 and 47 would have been outliers in FI at some point had the ensmbles been around at the time.

However as has already been said the UKMO and the ECM seem better at handling develoments once they come within a more reliable range. My quibble with the ECM is that in regards to winter synoptic situations is that can often flip flop at the closer range giving us all heart attacks just as everything looks nailed on only to revert to the nailed on scenario a run or two later.

Still I suppose it adds to the fun.

[Snowy Easterlies]

How can the UKMO be so different at day 6 compared to the GFS, whichever model is wrong it will lose its credibility for me, but then again the model which is wrong will probably be right another time so it will regain my trust we are all fickle

[Methuselah]

How can the UKMO be so different at day 6 compared to the GFS, whichever model is wrong it will lose its credibility for me, but then again the model which is wrong will probably be right another time so it will regain my trust we are all fickle

Unless of course, and perhaps more likely, they are both 'wrong'?

[snowking]

The model verification stats at day 6 are interesting. Looking at the slightly longer term picture. There is no doubt that simply due to its longer published runs the GFS is still king at picking up the trends in what we call FI. Quite often the past four or five winters cracking lone outlier has proved actually to be a tend setter and I have often remarked on here before that even the beginings of the winters of 63 and 47 would have been outliers in FI at some point had the ensmbles been around at the time.

However as has already been said the UKMO and the ECM seem better at handling develoments once they come within a more reliable range. My quibble with the ECM is that in regards to winter synoptic situations is that can often flip flop at the closer range giving us all heart attacks just as everything looks nailed on only to revert to the nailed on scenario a run or two later.

Still I suppose it adds to the fun.

Hmmmm....

http://www.emc.ncep....0_G2NHX_00Z.png

http://www.emc.ncep....H500mb_day8.png

Whilst I agree that the GFS first picked up on the trend this time around it is, in my opinion, a rarity of recent times.

It should also be noted, for anyone looking seriously at day 10 charts (aka FI), roughly 35% correlation. Theres a reason it is called fantasy island.

SK

Edited October 29, 2012 by snowking

[ribster]

Hmmmm....

http://www.emc.ncep....0_G2NHX_00Z.png

http://www.emc.ncep....H500mb_day8.png

Whilst I agree that the GFS first picked up on the trend this time around it is, in my opinion, a rarity of recent times.

It should also be noted, for anyone looking seriously at day 10 charts (aka FI), roughly 35% correlation. Theres a reason it is called fantasy island.

SK

I'm not sure that I follow, the weekend just gone was picked up way out by GFS (prior to the 228 chart that Steve M saved). ECM wouldn't have done as it simply doesn't go that far out. Can't argue with day 6, 7, 8 etc, ECM seems to be better, although from my limited understanding doesn't seem an awful lot. You appear to be advocating that we should only bother with 120/144, which would rather defeat the object of this thread for me. As has been said before, I can switch on the beeb for that kind of range. Obviously accuracy diminishes the further you go out, but the analysis and discussion of the possibilities in the medium to long term is what makes it so interesting/exciting.

[Gavin Hannah]

Well, not seen as many outputs recently, but from the ones I've seen today, 12Z esp, we are looking like being kept under a cool zonal flow. The 0C line more or less constantly covering the UK. A cooler than average start to November looks on the cards.

[mcweather]

Hmmmm....

http://www.emc.ncep....0_G2NHX_00Z.png

http://www.emc.ncep....H500mb_day8.png

Whilst I agree that the GFS first picked up on the trend this time around it is, in my opinion, a rarity of recent times.

It should also be noted, for anyone looking seriously at day 10 charts (aka FI), roughly 35% correlation. Theres a reason it is called fantasy island.

SK

35% correlation or not the fact is that nearly everything that eventually verifies starts somewhere out in FI its just a case of trying to pick out which scenarios they might be.

[snowking]

I'm not sure that I follow, the weekend just gone was picked up way out by GFS (prior to the 228 chart that Steve M saved). ECM wouldn't have done as it simply doesn't go that far out. Can't argue with day 6, 7, 8 etc, ECM seems to be better, although from my limited understanding doesn't seem an awful lot. You appear to be advocating that we should only bother with 120/144, which would rather defeat the object of this thread for me. As has been said before, I can switch on the beeb for that kind of range. Obviously accuracy diminishes the further you go out, but the analysis and discussion of the possibilities in the medium to long term is what makes it so interesting/exciting.

OK my friend,

I'm not saying that we shouldnt look that far, what I'm saying is that when looking that far we should be taking a monumental pinch of salt.

The event that has passed recently is an exception where the GFS did pick up a northerly flow at day 10 (shown below) from 12z:

However, even just 12 hours prior to this:

Already wrong with the flow direction by around 60-80 degrees

0z from the day before:

12z

So yes, within the day 10 period it did pick it up, this time. But it didn't consistently stick with it even at this range. Take a look at the 12z from day 9:

Or 12z from day 8:

So firstly, the GFS didnt really pick this up much beyond day 10, not with any certainty (or perhaps more relevantly, consistency). The ECMWF didnt pick up the eventual longwave pattern at day 10 in this instance, and hence me suggesting this was indeed an exception.

But the point I was trying to get across was that whilst statistics don't always tell the story, such a significant average over such a long period of time to me suggests that ECMWF should still be taken more seriously than GFS.

I think Steve has been incredibly intuitive in picking out certain patterns which are handled better by particular models, and again thats something that whole hemispherical statistics cant necessarily depict a perfect picture of.

But at the same time there are a lot of new members this year and I think its important to show that beyond 5-6 days, model accuracy is still very low indeed. I would also disagree with being able to use BBC television forecasts to see the weather 5-6 days in advance (with the exception of once a week on a sunday, which is of no use if you are at a wednesday and need to know the weather for sunday, for example) but thats a whole other discussion that we should discuss elsewhere

So i'm not saying we shouldn't look beyond 144 - i'm merely providing statistics to bear in mind when viewing charts at such range. I'd personally much rather use composites based around, for example GWO and MJO phasing, and find these generally more reliable at such a range (i'm aware that these outlooks are GFS and ECMWF derived, however there is far less variability and chaos potential than in standard NWP global output)

Kind Regards

SK

[snowking]

35% correlation or not the fact is that nearly everything that eventually verifies starts somewhere out in FI its just a case of trying to pick out which scenarios they might be.

Absolutely and theres no disputing that. However, that prognosis is fairly useless at such a range when you don't know what that scenario will be! And surely thats the point of forecasting? Picking an outcome rather than saying, well model X suggests 4 possible scenarios beyond day 10 at present, and so one of these 4 scenarios will come to pass.

Kind Regards

SK

Edited October 29, 2012 by snowking

[Cloud 10]

Whilst naefs shows a flattening of the mean pattern around Greenland in the 10 day area, the ridging returns for the end of week 2, we end the run with the three anomalies remaining - Greenland high, scandi low and Aleutian/Kamchatka high. Whilst we see low anomolys near the pole week 2, the spreads indicate that the core of these likely just north of Svalbard and the fact that the 850 anomolys stay positive throughout the arctic for the whole of week 2 says to me that there is no strong vortex forming. If the meridional pattern returns after an attempt to flatten,this to me is an encouraging sign for what is likely second half of November.

The 12z NAEFS continues the theme of those three anomalies,especially the one around Greenland,and if we compare the chart for now and the one at +300 hrs its a case of spot the

difference!

now.. much later...

No doubt subtle differences at the surface with regard to the UK's weather,but more or less a

back to square one scenario with cold and unsettled for mid-month.

Some good posts above by Snowking

Edited October 29, 2012 by Cloud 10

[SMU]

http://modeles.meteociel.fr/modeles/gfs/runs/2012102918/gfsnh-2-84.png?18

T84 & the GFS has the first welsh Gridpoint snow under T96 since the first cold wave at the weekend...

S

[ITSY]

GFS 18Z has temporary yet heavy transient snow in South Wales on Friday morning. Think it's highly unlikely, but a signal that with this cold zonality, much can happen at short notice

Edited October 29, 2012 by ITSY

[Glacier Point]

Indeed, the Kamchatka ridge is setting the hemispheric wavelength (arguably this is arising out of the SSTA in the North Pacific and mish-mash we've got which is neither El Nino nor La Nina). The signal in the SSTA in the North Atlantic are being enhanced by the current dip in angular momentum. I suspect a continiued below average trend in place for some considerable time, at least until the next westerly wind burst out of the tropics sets the MJO back towards a Svalbard block late month.

[Snow and storms]

The Zipping Friday low has been slightly upgraded on the 18z. This would produce a notable wind event and I would imagine a substantial amount of rain.

[damianslaw]

Its a classic amplified pattern at present - hurricane sandy is exceptionally large sticking in situ and being forced to move west into the USA eastern seaboard all courtesy of the king feature i.e. strong heights over Greenland, conversely we too are going to see our own large slow moving low pressure system - which is going to dominate things right through into the weekend.. this is a stuck amplified pattern quite unusual for the time of year which is traditionally one which sees a mobile atlantic pattern - but 2012 has been very unusual in this sense..

Beyond the reliable timeframe - both ECM and GFS are suggesting a toppling high pressure with a burst of westerlies thereafter before a renewed surge of heights over Greenland setting us up to a possible northerly. Its a very plausible outcome - thanks to the jet continuing to move on a more southerly trajectory.

Edited October 29, 2012 by damianslaw