Model Output Discussion:

[Cecil Hogmanay Melchett]

I think we have seen a dramtic flip overnight and the momentum seems to be for the trough to be much farther

west than though.The FAX chart was the 1st clue last night and ecm/ukmo are now suggesting the block will

put up more of a fight,yesterday ecm had us on the wrong side of any blocking but this morning it has flipped.

ukmo also looks better than yesterday with the trough filling and the block moving west.

6zgfs shows us what can go wrong though so lets wait until the 12z runs come out.

Id be delighted to see a repeat of the 0z ecm with the others coming on board.

[Paul]

Just a reminder - this is the model discussion thread, there are other threads available for chat about bbc forecasts etc.

[swilliam]

First time in a while we've seen mean height agreements between ECM/GFS...

http://www.meteo.psu...0z/hgtcomp.html

Ooooo.......don't you just love the thrill of it all

Yes the overall setup is very similar. However if you look closely you will see they are not quite the same. The ECM has the trough to the East of us whilst the GFS has the trough over us extending from the Atlantic. It is only a small difference but would make a big difference to the actual weather, the former give cold snowy conditions with the later probably near normal and rain..

[TEITS]

however given the UKMO has been the one model with the right trend over the last few days, although it was a little too progressive then we should view this as a decent middle ground solution for the timebeing between the ECM and the GFS.

Yes I agree the UKMO is the one to watch at the moment. The GFS in my opinion is even behind models such as GEM/NOGAPS after recent outputs. I didn't even bother looking at the 06Z after +144. However its rather ironic the latest BBC monthly forecast is suggesting what the GFS was saying a few days ago i.e blocking from Scandi to Greenland.

Based on the model outputs from the last week then the period of Feb 8th - 14th could see max temps either 15C or -4C.

[Cecil Hogmanay Melchett]

Yes I agree the UKMO is the one to watch at the moment. The GFS in my opinion is even behind models such as GEM/NOGAPS after recent outputs. I didn't even bother looking at the 06Z after +144. However its rather ironic the latest BBC monthly forecast is suggesting what the GFS was saying a few days ago i.e blocking from Scandi to Greenland.

Based on the model outputs from the last week then the period of Feb 8th - 14th could see max temps either 15C or -4C.

hah its frustrating teits but i happen to think we are now going in the right direction.

I think you'll be very happy with the 12z runs with ecm firming up on the block ridging towards Iceland and the Atlantic held at bay.

:cray:

[Bobby]

ECM is stunning. Big imrpovement on the ensembles too, the op is not a cold outlier at all, right on the mean. The control is even colder, though these are for Holland.

[grab my graupel]

Just a quick one from me before I disappear

As others have mentioned above, there appears to be some marked differences between the model runs this-morning with the ECM

sending a Shortwave underneath the block - while the GFS has low pressure sat to the West which keeps us in a kind of stalemate.

Therefore, until the earlier stages of the runs are sorted, anything thereafter should be taken with a pinch of salt.

The model differences can be cleary seen using the Netweather comparison charts:

http://www.netweather.tv/index.cgi?action=modelcomp;sess=

[snowmadchrisuk]

Interesting the MJO has stalled in phase 7 , the NAO is still forecast to go Negative and the AO is forecast to go very Negative maybe -5/-6 . This would fit in nicely with the BBC Monthly outlook and I fully expect the METO 6-15 forecast to continue with it's Colder later theme.

NAO :

http://www.cpc.noaa.gov/products/precip/CWlink/pna/nao_index_ensm.shtml

AO :

http://www.cpc.noaa.gov/products/precip/CWlink/pna/nao_index_ensm.shtml

[londonsnow]

well, is there much to add to the model discussion. i don't remember such fluctuation in many years of model watching. i think a less cold scenario is nailed for especially the south this week, but its beyond that which is in the balance. looking at the beebs monthly update they are going for an easterly blast of some considerable impact, but with FI hovering so much closer than normal i remain sceptical.

[Iceberg]

I've not posted much over the last 36 hrs, essentially as I still believed my original comment on Saturday of don't trust the models and if you don't trust the models then there isn't much point talking about them.

A quick comment about the UKMET GM performance, Yes the UKGM was right to close the door to the Se movement of energy in the atlantic, thereby undercutting the jet and giving us the northerly that was progged on Saturday.

However it was crucially wrong by putting as much energy above the jet as it was showing on Saturday. So we have not been put into a no win situation.

IMO the models have started to come together, more energy in the atlantic pre 144, still little sign of a pressure release value to the SE, which keeps the block for the time being just out of reach.

However because UKMET was right about the med pressure not falling as much as was progged by GFS and ECM does not mean it will be right next time.

The key for me is what happens down south, with the Azores hign and the med/spanish high.

The 00Z ECM was towards the lower end of the EPS for London, it certianly wasn't without support though. At 200+ there really is no tight clustering on the EPS, so as the saying says anything goes.

[Guest North Sea Snow Convection]

The bbc monthly out look updated today predicts an easterly!!!!

Woken up to snow again!! and this today brings us to our total of 7 snow events this winter, this never happens in Sussex (average is normally 1-2 snow events per winter for the last 10+ years!) - so this is a winter thats got so much bite to it, it will be back again after the mild spell !!

Quite a winter isn't it! And, aww, I love your bunny rabbit avatar!

The METO were saying over the weekend that they had a lot of confidence in data for the extended 10 to 15 day period and that the uncertainty was around the middle 4 to 7 day area, in terms of the extent of the atlantic energy moving eastwards from the mid atlantic low pressure.

It is now clear that the UKMO picked up the right trend for the breakdown to atlantic weather expected this week, and that the other models (GFS in particular) were being too progressive in trying to bring in the easterly.

However, with that METO assessment in mind, it is also becoming clearer that the UKMO also initially was the first to overstate that atlantic energy, showing it extending virtually into Scandinavia and effectively unseating the -AO pattern. I think I commented at the time that it was very unlikely. However, GFS was last to pick up the breakdown signal for this week, and now, might be the last model to pick up the signal for the possible easterly from the weekend/start of next week.

This morning, although there are slight differences in the exact pressure detail, it is not too incorrect to suggest that both ECM and UKMO are going the same road in terms of the atlantic retreating and in the case of ECM, a clear indication of the actual easterly itself evolving as the residual energy in the atlantic is sent SE 'wards under the westwards moving block over Scandinavia and Iceland/Greenland.

The ECM last night suggested an undercut of energy might occur at the t96 tineframe, and it goes to show that it needs to come within a 3 to 4 day timeframe in these situations before the right trend is picked up. I say the right trend, but of course we still need full agreement from all the models. However, with the assessment in mind from the METO, and the way the modelling is pulling right back from the atlantic invasion, it does appear as though we might this time be on a steadier road towards one solution.

Not there yet though, but hopefully steady as she goes

Edited February 1, 2010 by North Sea Snow Convection

[snowmadchrisuk]

I've not posted much over the last 36 hrs, essentially as I still believed my original comment on Saturday of don't trust the models and if you don't trust the models then there isn't much point talking about them.

A quick comment about the UKMET GM performance, Yes the UKGM was right to close the door to the Se movement of energy in the atlantic, thereby undercutting the jet and giving us the northerly that was progged on Saturday.

However it was crucially wrong by putting as much energy above the jet as it was showing on Saturday. So we have not been put into a no win situation.

IMO the models have started to come together, more energy in the atlantic pre 144, still little sign of a pressure release value to the SE, which keeps the block for the time being just out of reach.

However because UKMET was right about the med pressure not falling as much as was progged by GFS and ECM does not mean it will be right next time.

The key for me is what happens down south, with the Azores hign and the med/spanish high.

The 00Z ECM was towards the lower end of the EPS for London, it certianly wasn't without support though. At 200+ there really is no tight clustering on the EPS, so as the saying says anything goes.

I really think you have to follow Instinct at the moment , and I'm pretty sure that's what the senior Forecasters are currently doing . The background signals are all set for a big freeze with Negative Nao , AO and of course the MJO stalled in a very good place . We have a Huge Block building over Scandinavia again and constant signs of High pressure moving South from the Arctic and Into Greenland . The 00z Ensembles were pretty good with the Mean at -5 or below after the 8th . The 6z GFS Ensembles have flipped and are not as good but I don't think I am alone by saying I lost all faith in the longer range Ensembles along time ago , they flip from one extreme to another way to often , quite often from run to run . The Control run has been very cold lately , usually Colder than the Operational and I understand the control is the only other Ensemble member that is run at a slightly higher resolution .

[Methuselah]

well, is there much to add to the model discussion. i don't remember such fluctuation in many years of model watching. i think a less cold scenario is nailed for especially the south this week, but its beyond that which is in the balance. looking at the beebs monthly update they are going for an easterly blast of some considerable impact, but with FI hovering so much closer than normal i remain sceptical.

Maybe a tad OT. But I hope not??? :unsure:

Being of a particular vintage (old fart! ) I can recall all the uncertainty in the 1960s' forecasts...

For the past 20 years, there's been almost no chance whatever of a Beast From The East - successive Euro-Highs have kept the near Continent too warm and snow-free??? And, any Northerlies were about 50 miles wide (exaggeration!) and flew into Europe...

This time, in contrast, there a broad swathes of cold air to our NW, N and NE - all of which are resisting the Atlantic...No wonder then, that the models are having a hard time!!

[full_frontal_occlusion]

Worth mentioning that Feb 7th to 14th is also known as one of the Buchan spells. I think the next one is during April.

Interesting you mention Buchan - we very often forget (or at least most do not know) of 'rules of thumb' postulates with some surprising statistics for the British Isles, long before NWP;

Brookes C.E.P (1889 - 1940) weather 'singularities'(rates of change = 0 in other words static) which accordingly predict:

mid Jan (18 -24th) anticyclonic high pressure 87% probablity

late January (25 - 31st) cyclonic stormy 85%

early Feb (8 - 16th) anticyclonic 56%

late Feb (21 - 25th) cold spell 42%

late Feb early March (26 - 9th) stormy 88%

Lamb Hubert (1913 - 1997) natural seasons:

LATE WINTER (20 Jan - 29 March)

lengthy spells but no preference between zonal-mild-windy & blocked/cold. i.e. 50% either

Coldest winters when persistent blocking highs Scandinavia/Iceland.

Polar Continental / Arctic Maritime ... cold. northerly types 5 days max.

20th - 23rd January Anticyclonic EUROPE + S. & E. Britain

27th January - 3rd February renewed storminess gales and rain or snow

Lows pass into North & Central Europe from Atlantic. The first lows of the series commonly approach England from the South-West (N.B. liability for freezing rain with advance of mild air after some days of frost). Anticyclonic type rare.

8th - 13th February Anticyclones

Record frosts in cold winters. Fogs common inland. Southerly & Easterly types common in Britain.

26th February - 9th March Cold Stormy period

Northerly outbreak from Norwegian Sea and cyclonic weather over N. Sea and surrounding lands. Cyclonic type maximum 26th February - 2nd March; Northerly type maximum 28th February - 3rd March.

Buchan Alexander (1829 - 1907) Spells:

7th - 14th February: COLD

11th - 14 April: COLD

Barry & Perry:

20 - 23rd January - Precipitation minimum in central / Southern England >70%.

So, combining these rules we should expect:

Mid - late Jan; Anticyclonic dry and frosty.

Late Jan - Early Feb; A stormy Atlantic period with lows approaching from the southwest and passing into north and central Europe with a liability for freezing rain.

Early - Mid Feb; Anticyclonic (Southerlies and Easterlies) with potential for record frosts.

Late Feb - Cold Spell with northerly outbreak from Norwegian Sea.

Interesting indeed.

ffO.

Edited February 1, 2010 by full_frontal_occlusion

[Just Below Zero]

I really think you have to follow Instinct at the moment , and I'm pretty sure that's what the senior Forecasters are currently doing . The background signals are all set for a big freeze with Negative Nao , AO and of course the MJO stalled in a very good place . We have a Huge Block building over Scandinavia again and constant signs of High pressure moving South from the Arctic and Into Greenland . The 00z Ensembles were pretty good with the Mean at -5 or below after the 8th . The 6z GFS Ensembles have flipped and are not as good but I don't think I am alone by saying I lost all faith in the longer range Ensembles along time ago , they flip from one extreme to another way to often , quite often from run to run . The Control run has been very cold lately , usually Colder than the Operational and I understand the control is the only other Ensemble member that is run at a slightly higher resolution .

A lot of talk lately about the reliability of the ensembles and I can see where people are coming from.

However we must remember that the ensembles are only tweaked at small variances from the early part of the operational. Therefore if a model's operational run has changed markedly early in its run to a milder theme say, then it follows that all the ensembles will start from that milder base. This will have the effect of the flip flops we have seen recently.

The ensembles turn arounds are therefore more likely to occur when Model divergence starts early say T72 instead of T120.

With this in mind the use of ensembles can still be used as a useful tool :unsure:

[SP1986]

I do get the feeling that the models are going to revert back to some kind of easterly, but this doesn't necessarily mean it will occur, and it could cause a lot of unwanted stress, but the ECM is actually something we should be watching now I suspect, although expect a game of cat and mouse.

ps you can tell the GFS is struggling with this more than any other model by taking a look at its surface temperature values over wide areas.

[Iceberg]

The difference in the GFS normal and alternative ENS is quite staggering, with much more uncertainty with the alternative(new) ENS.

We still need to wait 24hrs or so IMO, before we even enter a discussion on how an easterly will occur.

Ensembles are still important(particularly the ECM EPS for London), however I couldn't put up any models output and say it's got a good handle on things.

[southbank]

I do get the feeling that the models are going to revert back to some kind of easterly, but this doesn't necessarily mean it will occur, and it could cause a lot of unwanted stress, but the ECM is actually something we should be watching now I suspect, although expect a game of cat and mouse.

ps you can tell the GFS is struggling with this more than any other model by taking a look at its surface temperature values over wide areas.

I think you are right , I think we are still in for some wild swings as witness towards the end of last week and the flip flopping over the weekend . Unless the teleconnections stop supporting a signal for a northern block to form then despite the gloom on here after sunday runs there is still chance of a colder outbreak . I feel next week in the later stages is where we will start firming up on a cold soultion however thats 216 hrs away and not one model can cope at the mo at that distance with blocking signals and the energy in the jet as it exits the US so the rollercoaster to last a bit longer methinks

[bluearmy]

I really think you have to follow Instinct at the moment , and I'm pretty sure that's what the senior Forecasters are currently doing . The background signals are all set for a big freeze with Negative Nao , AO and of course the MJO stalled in a very good place . We have a Huge Block building over Scandinavia again and constant signs of High pressure moving South from the Arctic and Into Greenland . The 00z Ensembles were pretty good with the Mean at -5 or below after the 8th . The 6z GFS Ensembles have flipped and are not as good but I don't think I am alone by saying I lost all faith in the longer range Ensembles along time ago , they flip from one extreme to another way to often , quite often from run to run . The Control run has been very cold lately , usually Colder than the Operational and I understand the control is the only other Ensemble member that is run at a slightly higher resolution .

The control is run with the same starting data as the op but at the same res as the other ens members.

the parallel ens are better than the operational ens on the 06z with a split emerging and the colder members a lot colder and the mean down to -6c/-8c by day 9. the mean flow becomes se by T180 with an mean undercutting trough looking the likely scenario. given that many of the members still show higher uppers, if the trend is to cold, the ensemble mean is likely to flip to very cold.

[Duane S.]

The control is run with the same starting data as the op but at the same res as the other ens members.

the parallel ens are better than the operational ens on the 06z with a split emerging and the colder members a lot colder and the mean down to -6c/-8c by day 9. the mean flow becomes se by T180 with an mean undercutting trough looking the likely scenario. given that many of the members still show higher uppers, if the trend is to cold, the ensemble mean is likely to flip to very cold.

Hi Blue Army,

And a nice control run too again! :lol:

Do you know what exactly is tweaked each time between all the members?

Is it always the same tweaks to the same members?

Also, why not just run all of them (with tweaked initial data on all bar the op) on a higher resolution and get shot of the control? Why have a lower resolution at all?

Never quite fully got my head around the ensembles.

[stewfox]

If we start with the big picture the deep cold pool over Europe starts pushing east and this meets and this is the important bit ,the very cold pool of air over Russia amd that starts moving west once again ,this will be the trigger or mechanism which will spill the cold air once again to our shores,

Like a balloon effect , squeeze it in, it pops back .

You can see how the cold air is being pushed back even Moscow -1c today (max) but -12c (max) Friday.

You can also see that balloon being squashed here

http://uk.weather.com/global

Edited February 1, 2010 by stewfox

[Nick F]

Also, why not just run all of them (with tweaked initial data on all bar the op) on a higher resolution and get shot of the control? Why have a lower resolution at all?

Never quite fully got my head around the ensembles.

I imagine it takes alot of data number crunching to run a high resolution run, remember they are global models not just covering Europe. So if you ran 20 high res runs, or in the case of ECM, 50 high res runs, thats an awful lot of computing power needed. So, I think that is why the ensembles are run at lower res, due to computing constraints as mentioned.

With regards to the outlook into the medium range, we lose the meridional flow by mid-week this week as strong and flat jet comes tearing across the Atlantic towards NW Europe, we need to see this jet weaken and amplify or dive further south if we want to see the cold air to the east back west. Signs are the upper trough over the NW Atlantic being supported by cold air spilling off the PV over NE Canada will slowly weaken and disrupt with the jet edging further south, so just need to ride out hopefully a short period of a stronger Atlantic jet before the N Atlantic flow finally weakens. Though we still need blocking to build and position favourably to bring cold our way. Certainly no reason yet to give up on a return of a colder and snowier outlook for later on.

[snowmadchrisuk]

The control is run with the same starting data as the op but at the same res as the other ens members.

the parallel ens are better than the operational ens on the 06z with a split emerging and the colder members a lot colder and the mean down to -6c/-8c by day 9. the mean flow becomes se by T180 with an mean undercutting trough looking the likely scenario. given that many of the members still show higher uppers, if the trend is to cold, the ensemble mean is likely to flip to very cold.

The Control run varies depending on which run your viewing . Some General Information from the Noaa is below :

There will always be a trade-off between the resolution at which the forecasts are made and the number of ensemble forecast members, due to limited computational resources. Since the impact of using a higher resolution model is not detectable with traditional skill scores beyond a few days (Tracton and Kalnay, 1993), at NCEP we truncate the resolution of the nominal MRF and AVN runs (T126 truncation, ~100 km) to T62 (~200 km) at a lead times of 7 and 3 days at 00Z and 12Z, respectively. At 00Z there is also a "control" totally T62 run. In addition to this control forecast, 10 forecasts with T62 resolution are run from 00Z starting from slightly perturbed initial conditions. At 12Z four additional forecasts are generated from perturbed initial analyses. Hence, there is a total of 17 individual global predictions generated daily. All forecasts are run to 16 days with the latest version of the EMC MRF global model (Kanamitsu et al., 1993). Evaluations indicate that for the first week or so for daily forecasts, the set of 17 ensemble members is sufficient. For the currently operational 6-10 day mean outlooks (and future "week2" forecasts), additional information is added by including the runs from up to 48 hours prior to "todays" 00Z set in the time (or lagged) average sense (Hoffman and Kalnay, 198*). In this application the total the number of forecasts (ensemble size) equals 46.

[snowmadchrisuk]

Also some more very interesting Information from the Noaa , is them explaining how the ensembles can sometimes all go for the wrong solution .

Finally, keep in mind that there is no guarantee that the above methodology "finds" all the possible growing modes or, equivalently, the ensemble will reliably encompass all possible outcomes in every situation: we cannot run enough perturbed forecasts (with, for example, different initial perturbation sizes) to populate the whole forecast distribution all the time. Moreover, remember that the forecast model is not "perfect", and model error, as well as initial condition uncertainty, will contribute to the distribution of predictions within the ensemble (especially systematic errors which may drive all the solutions in the same - wrong - direction). Overall, however, verifications indicate that the ensemble system as now constructed does provide enhanced skill through ensemble averaging and usefully reliable probability estimates.

[The PIT]

Also some more very interesting Information from the Noaa , is them explaining how the ensembles can sometimes all go for the wrong solution .

Finally, keep in mind that there is no guarantee that the above methodology "finds" all the possible growing modes or, equivalently, the ensemble will reliably encompass all possible outcomes in every situation: we cannot run enough perturbed forecasts (with, for example, different initial perturbation sizes) to populate the whole forecast distribution all the time. Moreover, remember that the forecast model is not "perfect", and model error, as well as initial condition uncertainty, will contribute to the distribution of predictions within the ensemble (especially systematic errors which may drive all the solutions in the same - wrong - direction). Overall, however, verifications indicate that the ensemble system as now constructed does provide enhanced skill through ensemble averaging and usefully reliable probability estimates.

Good job they're not perfect otherwise we'd have nothing to discuss. No more ramping no more baiting no more phantom bartletts etc etc...

Have you got a link to noaa about the assemblies would be an interesting read.