Model Output Discussion - The Start Of Winter

[warrenb]

To be honest can you stop using the Mean, as it is irrelevant with the data being used. Ensemble stamps are the best as you can see if different scenarios are being seen or a trend. The mean is pointless.

[ukpaul]

London GEFS Ensembles, looks like the strongest cluster of runs with below -5 uppers for weeks.

 

 

Edit: and I agree, long term means are useless in isolation. Posted side by side from a few timeframes to get a sense of movement then they might show something. Anomaly charts at that range at least show where we are in comparison to the norm.

Edited December 16, 2014 by ukpaul

[The Post-modern Winter]

London GEFS Ensembles, looks like the strongest cluster of runs with below -5 uppers for weeks.

 

 

Edit: and I agree, long term means are useless in isolation. Posted side by side from a few timeframes to get a sense of movement then they might show something. Anomaly charts at that range at least show where we are in comparison to the norm.

 

I agree regarding the mean; even from a fundamentally mathematical point of view I can't see it being of any use initially in the scenarios that many are searching for, i.e. a pattern change. Can anybody more experienced tell me how the mean charts would have looked two weeks before the onset of the late Nov' 2010 cold for instance, or any other relevant cold spell? 

Edited December 16, 2014 by The Post-modern Winter

[Barry12]

IDO, I remember Ian Fergusson saying the Met Office use the operational at longer range rather than the mean.It's looking like the Christmas period will be cold and wintry, numerous runs now have shown this. A cold Boxing Day has been shown for days on the models, yes it may only be temporary but it's looking likely now. The mean will always be behind at that range, as any signal will be muted as Steve Murr has said many times.

Edited December 16, 2014 by Barry95

[mountain shadow]

To be honest can you stop using the Mean, as it is irrelevant with the data being used. Ensemble stamps are the best as you can see if different scenarios are being seen or a trend. The mean is pointless.

Agreed Warren, IDO's continued use of the mean at t+million is irrelevant given the vast scatter the further out you get. Pointless.

[bluearmy]

Barry - Exeter do NOT use the operational ahead of the ens guidance beyond a week.

They will use the op if it is in good agreement with the ens but I struggle to think of instances where an op out of kilter with the ens after a week would be preferred guidance.

The ens mean is fine to use as a tool, as long as it's done in conjunction wth the spreads, clusters and anomolys.

I will post again re the potential cold post xmas after the 12z ens are out. Conflicting signals emerging re short and sharp or something more sustained.

[IDO]

To be honest can you stop using the Mean, as it is irrelevant with the data being used. Ensemble stamps are the best as you can see if different scenarios are being seen or a trend. The mean is pointless.

 

The GFS P has uppers 9c below the mean on Christmas Day. It is the coldest solution and statistically it is an outlier assuming the standard deviation is +/- 40% each way. So even looking at the clusters it has nominal support. At that time scale the Mean is usually showing around 33-50% better verification than the op.

 

 

IDO, I remember Ian Fergusson saying the Met Office use the operational at longer range rather than the mean.It's looking like the Christmas period will be cold and wintry, numerous runs now have shown this. A cold Boxing Day has been shown for days on the models, yes it may only be temporary but it's looking likely now. The mean will always be behind at that range, as any signal will be muted as Steve Murr has said many times.

 

Hi Barry. I believe that the GFS P is the shape of things to come. However not till January. Using the OP after D5 would not be my idea of getting good forecasts; you only have to look at ECM D8-10 charts to realise that. I always use the mean after D7 and look for clusters within that mean. You will be continually disappointed if you rely on UK cold based on D10 operational charts. Of course only my opinion and everyone has their own ideas.

[mushymanrob]

So in 10 years you have yet to come to the understanding that people will look for cold weather in winter?   

Long range models and other sources are indicating a pattern change, I think it's perfectly reasonable to look for any sign of this change in the models, you are of course free to discuss what you like, not sure why your interested in talking about other peoples interest 

 

hi

...erm, thats not quite what i meant, of course i understand that seasonal people look for seasonal weather that interests them. 

but what i dont understand is the jumping to premature conclusions, or conclusions made on odd favoured fi runs . ive seen posts praising fi charts as the herald of something different emerging, only to be dashed later on. so yes, by all means note interesting developments but dont crack open any champagne until something concrete becomes apparent in a reliable timeframe..

and in summer i adhere to this too when heatwaves are shown in fi, im just suggesting abit of sobriety and open mindedness, otherwise the likelyhood of great disappointment is pretty large.

 

An odd post Mush..All part of model discussion, As im sure all members are aware that looking further into the runs are a little speculative whether showing mild or cold. Most are looking at small changes in upstream signals and makes for interesting viewing and lots to learn for members which is what this thread is all about. I would of thought after 10 yrs you would have been wiser to this.

 

just trying to add abit of objectivity

 

If  you are looking at an operational run in isolation, yes, but it's perfectly legitimate to get broad clues from ensembles and their means out at those ranges for a pattern change IF they are continually showing that change.

 

There has been a pattern change consistently been flagged upstream over NE Pacific and N America by GEFS and EPS.

 

This pattern change takes place way upstream with signs of the N Pacific jet stream retracting and the Aleutian low retrogressing in the means by day 10, thus leading to a ridge pushing north across Alaska and Yukon towards the pole ... ridging north over Siberia too perhaps leading to cross polar HGT rises. Downstream, a trough is forced south over central Canada and US Plains intially before transfering to eastern N America by day 10. This forces ridging north over the Davis Strait and Greenland.

 

ecmwf-ens_z500a_nhem_11.png

 

This ridging in the far NW Atlantic is not looking to form a block with current guidance - as John H and others suggest ... but may hang on enough to drop some deep cold south across the UK around 26-29th. Low heights are still going to be a problem to shift over Nern Canada and Greenland, so any ridge north over the NW Atlantic will struggle to form a block, though there are signs of the AO going -ve later this month:

 

ao.sprd2.gif

 

Vertical wave propagation has been rather weak of late given the zonal nature of the flow in the northern hemisphere, but with these changes upstream and signs of wave number one kicking in, we could see more blocking reappear as we had into the New Year, all this without signs of the potential SSW yet, of which there appears an increased chance of in Jan.

 

hi nick, appreciate that detailed analysis.

yes i am looking at the ops in isolation, i dont pretend to understand the longer term signals so ill stick with what i understand. wouldnt longer term signals be more likely to be inaccurate? and will manifest in the ops before becoming reality. so ill stick with the forseeable.

so the signs are for an initial ridge to our west post crimbo, which is expected to be transitory (as the gfs currently suggests) , before a second ridging event (for want of a better description) is a more decisive blocking feature in the new year?

this'll be interesting!

[Carlrg]

Run after run after run, all I can see is a very transient set of weather. Not boring by any means but even deep into FI, personally I can see no entrenched cold coming to our shores.

Edited December 16, 2014 by Carlrg

[Smiler1709]

Growing trends on the GEFS that temps are gonna start to fall after Xmas, there's more runs now starting to show with hints of Northern blocking or blasts of cold air from the artic. Obviously there still is the milder runs but it's certainly encouraging to see stone bone chilling runs in there also.

I think the big change is gonna come in January when I think our first real countrywide cold spell will hit. So far this winter has been pretty awesome already compared to last years rubbish, we have already had a good ten frosts and some really decent chilly winter days.

So already this winter is completely different, colder, Atlantic is no were near as powerful and with the continued model trends for colder weather from the northwest instead of horrible westerly winds of last year.

Mild blip this week then something more seasonal for the festive period. Plenty to be happy about.

Growing trends on the GEFS that temps are gonna start to fall after Xmas, there's more runs now starting to show with hints of Northern blocking or blasts of cold air from the artic. Obviously there still is the milder runs but it's certainly encouraging to see stone bone chilling runs in there

Mild blip this week then something more seasonal for the festive period. Plenty to be happy about.

Edited December 16, 2014 by Smiler1709

[No Balls Like Snow Balls]

Albeit run at a lower resolution than the ECM DET the below ECMWF 500MB EPS control for the 8-15 day period shows a Positive anomaly over Greenland and a negative anomaly over Scandinavia.

Make of that what you will :-)

[mountain shadow]

Run after run after run, all I can see is a very transient set of weather. Not boring by any means but even deep into FI, personally I can see no entrenched cold coming to our shores.

You don't have to look very far to find it...

 

Latest GFS P..

 

http://hib.nwstatic.co.uk/gfsimagesnew/gfs.20141216/06/384/h850t850eu.png

 

NO forecast

 

http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao_index_ensm.shtml

 

NAO forecast

 

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

 

Iceland ensembles almost universally agree on pressure rises to between 1010mb and 1030mb

 

and finally the height anomalys at 10 days clearly showing the potential for a cold spell..

 

http://hib.nwstatic.co.uk/gfsimages/gfs.20141216/06/240/hgt500-anomaly.png

 

There is a clear potential there and found these links in about a minute, so to say you can't see a cold spell has no basis in the current model output.

 

Edited December 16, 2014 by mountain shadow

[Smiler1709]

A few of us tripping out on the post button lol

[Carlrg]

Thanks for that Mountain Shadow.  

Edited December 16, 2014 by Carlrg

[johnholmes]

I agree regarding the mean; even from a fundamentally mathematical point of view I can't see it being of any use initially in the scenarios that many are searching for, i.e. a pattern change. Can anybody more experienced tell me how the mean charts would have looked two weeks before the onset of the late Nov' 2010 cold for instance, or any other relevant cold spell? 

 

i will see if I can dig out the research I was doing at the time on the 3 anomaly charts I was checking and still use. From memory they well predicted about 2 weeks or so that the cold would arrive when it did. May be wrong but will try and find them and post in a pdf format in here this evening.

[johnholmes]

hi ba

re this

Barry - Exeter do NOT use the operational ahead of the ens guidance beyond a week

It sounds reasonable but do you know from asking them or is this an assumption on your part?

[Polar Maritime]

Barry did say Ian mentioned it John.

Edited December 16, 2014 by Polar Maritime

[Frosty.]

Very happy to see the models firming up on a cold blast around the middle of next week which coincides perfectly with Christmas. Not only that, the MO also talking about a spell of cold showery weather during this timeframe so the chances of some of us seeing snow on the big day have gone up a notch or three. Looking at the overall picture, we have a milder few days to endure but it will then turn colder and showery from the NW by Friday into the weekend with snow on hills and frosty nights..then milder, wet and windy early next week before the cold plunge and then beyond that it's more of the same but given the background signals, I'm hopeful we will see a marked pattern change to a colder blocked Jan / Feb :-)

[IDO]

hi ba

re this

Barry - Exeter do NOT use the operational ahead of the ens guidance beyond a week

It sounds reasonable but do you know from asking them or is this an assumption on your part?

 

Hi. I always thought that the UKMO only goes out to T144. They use ECMWF and MOGREPS for LR or so I thought. I am sure Ian F will correct me if I am wrong.

 

http://www.metoffice.gov.uk/research/modelling-systems/unified-model/weather-forecasting

[No Balls Like Snow Balls]

Hi. I always thought that the UKMO only goes out to T144. They use ECMWF and MOGREPS for LR or so I thought. I am sure Ian F will correct me if I am wrong.

 

http://www.metoffice.gov.uk/research/modelling-systems/unified-model/weather-forecasting

I am pretty sure Ian did say the other day not to use the mean charts as they are skewed by the historical climatological averages.

[IDO]

I am pretty sure Ian did say the other day not to use the mean charts as they are skewed by the historical climatological averages.

 

Hi. That's the anomaly charts.

[bluearmy]

ian talked about not using the ens mean into the distance as it cannot illustrate the likely verified chart due to too much smoothing out from the  51 members. he was speaking about latter week 2 and ecm 32 day ens data. he commented that exeter will use the clusters to get a handle on what is more probable at that range. clearly, we require the anomolys to give us guidance where clusters are not availble and spreads where they are

 

talk about chinese whispers - send three and fourpence - i'm going to a dance' !!

[Downpour]

Enjoying this winter thus far with some bright, sunny, cold days and plenty of frosts out here on the edge of London by Epping Forest. 

 

Models have now been hinting at a more interesting period during the Christmas holiday season, and this trend continues today. 

[knocker]

Well I wouldn't have thought thr GEFS was hinting at a cold 'spell' over Xmas just your bog standard NW in the north and westerly in south. Developments further afield but  where oh where are the Arctic incursions I ask myself.

Charts weatherbell

Edited December 16, 2014 by knocker

[Nick F]

hi

...erm, thats not quite what i meant, of course i understand that seasonal people look for seasonal weather that interests them.

but what i dont understand is the jumping to premature conclusions, or conclusions made on odd favoured fi runs . ive seen posts praising fi charts as the herald of something different emerging, only to be dashed later on. so yes, by all means note interesting developments but dont crack open any champagne until something concrete becomes apparent in a reliable timeframe..

and in summer i adhere to this too when heatwaves are shown in fi, im just suggesting abit of sobriety and open mindedness, otherwise the likelyhood of great disappointment is pretty large.

 

 

just trying to add abit of objectivity

 

 

hi nick, appreciate that detailed analysis.

yes i am looking at the ops in isolation, i dont pretend to understand the longer term signals so ill stick with what i understand. wouldnt longer term signals be more likely to be inaccurate? and will manifest in the ops before becoming reality. so ill stick with the forseeable.

 

 

The high res operational runs are best for detail up to a week, but beyond that, ensemble forecasting will typically be a more reliable indicator than a deterministic run of a model. Trouble with a deterministic/operational run is that new small features picked up that weren't on a previous deter run can skew it in a particular direction  - which may not be the correct direction!. By running pertubations from the set of intial conditions of the state of the current atmosphere, we gain more of an insight into the uncertainty of a model forecast. So this where ensemble guidance comes in use after a week.

 

Admittedly, the trouble with using an ensemble mean forecast, is that averaging out differences of the ens members, as the model forecast progresses, can leading to 'smoothing' out of any small scale features, leading to uncertainty. So using the ensemble mean on its own is not always an accurate guide, rather the ens mean can be supplemented by ens clusters that form (groups of ens members with similar evolutions). Some days the forecasts from ensemble members can be similar, other days they can be quite different from each other, with alot of spread - this is when caution is required in using the ensemble mean as guidance. Rather look at all ensemble members, time-consuming as that maybe.

 

Unfortunately, we don't have access to the EPS members beyond T+120 now, so as bluearmy mentions, we rely on the EPS H500 mean/anomaly charts - to get a broadscale idea of H500 pattern changes , to day 10, but mainly when they are persistently showing that change. GEFS we obviously have access to all the run, but with only 20 members, it's difficult to get meaningful clusters to compliment the ENS mean.

Edited December 16, 2014 by Nick F