General Model Output Discussion

[nick sussex]

Morning Dave, you're right of course when you say that the point of the thread is to discuss what the charts show, but its also about analysing how likely it is that what the charts are showing will verify, or how close to verifying they are likely to be. It should not be about just shamelessly ramping them and leaving the less knowledgeable to believe what they like. I have to say that one or two members, who should know better, have let their excitement get the better of them and consequently let their objectivity slip. Yes these are the best winter synoptics we have seen but that's not a reason to abandon caution and cool analysis. Its been a good start to winter, I've already seen more days with snow falling than I've seen for a good few years and some cold minima's, however it is still not yet what I would call a truly remarkable winter, that will hopefully cement itself over the next few weeks. Just watching the 06z coming out now, a run which I suspect will dent some confidence in a prolonged easterly, and although I would say the GFS 06z is the poorest performer of the four GFS runs, it does show another alternative evolution a rather dryer one as well.

I think we have to accept the fluidity of the pattern in that the pattern is constantly evolving, the gfs 06 hrs run IMO has the right idea but then doesn't follow through with the most likely scenario, the high should transfer towards Iceland and then a trough develop over Scandi, the troughing over the eastern USA in line with a very positive PNA should be aligned north south with WAA building pressure over Greenland, its a reload until the upstream pattern changes. We should then see another northerly before yet again the reload back to easterly.

I would be very surprised if the ensembles don't have quite a few members with those synoptics.

[bluearmy]

the closer attendance of the block in FI GFS fits in sell with yesterdays METO 15 dayer. as long as we get decent snowfall nationwide before then, i guess most will take this option in order to prolong the cold spell and provide for an easy reload to a greeny block later on.

[weather eater]

This is what GFS 06z op run last Sunday was forecasting for this Sunday

And from a fortnight ago for this Sunday

I can't say I'm surprised, it is as I said IMO the poorest performer of the four GFS runs. If the 18z is the pub run, the 06z is the, lord I’ve woken up with a terrible hangover and can’t think straight run.

[johnholmes]

I think we have to accept the fluidity of the pattern in that the pattern is constantly evolving, the gfs 06 hrs run IMO has the right idea but then doesn't follow through with the most likely scenario, the high should transfer towards Iceland and then a trough develop over Scandi, the troughing over the eastern USA in line with a very positive PNA should be aligned north south with WAA building pressure over Greenland, its a reload until the upstream pattern changes. We should then see another northerly before yet again the reload back to easterly.

I would be very surprised if the ensembles don't have quite a few members with those synoptics.

when you say SHOULD Nick, do you mean scientifically as the programme says or or you more at hopecasting perhaps?

Of course you could be right and we both have to wait until that time to see whether you or the model is correct. Personally what the model is showing seems reasonable to me but then I tend not to follow each nuance of every run, especially at long time frames.

As to the ensembles-the whole idea of that is to step change the start data to see how many variations within a reasonable scale the model will then evolve. This then allows fairly complex statistics to be done an the likely % of any scenario being played out. All very complex and a bit beyond me I'm afraid. However, the ensemble idea when transferred to basic forecasting for a customer does allow degrees of probability for any particular forecast scenario to be given to the customer.

[Duane S.]

Can I bank the 06z, please?? The first chart within the reliable(very 'ish') timeframe that gives ME potential of a good dumping. My first snowflake of the winter would be followed in very quick succession by many others if this verifies!

This has been hinted at for a while on the charts and seems a pluasable outcome. As the front drops down through us it is heading towards where the energy is, a formation of a small trough before it gets absorbed into the parent Med one, why not? Wishful thinking maybe but this is the first time this winter I have been able to see an evolution within 4 days that COULD give me down here a bit of snow.

[Airedalejoe]

why, what is the matter with that slack spell?

If you look at the T+120 Fax it shows the 528 line on summer holiday in Iberia-the 850 thickness is on much the same plane.

Neil STOP worrying-EVEN if mild air arrived post T+120 which it will not you would still have had almost a week of bitter cold?

For anyone interested in the science of showers off the North Sea-why the Met O probably have heavy snow showers in some of their central and eastern forecasts then perhaps a trip to my last blog may help.

Its about skew-t's this week along with mean temps for this week for my area-a fair example of low level weather for parts of eastern England.

I'm on the point of updating it with comments using the 00z.

enjoy

In laymans terms, the North Sea can be an enormous rain/snow generator for my area. Can make large snowfall totals in winter with what I guess is what is known as 'lake effect' snow in North America.

In summer heavy rain can fall with high presure very close by with other areas being warm and sunny!

Annoying in the summer but fantastic at times like this. Another 8cms last night. It's a winter wonderland here.

[johnholmes]

This is what GFS 06z op run last Sunday was forecasting for this Sunday

And from a fortnight ago for this Sunday

I like that Mr D-illustrates just how fickle a computer can be-almost makes an old met man happy!

[weather eater]

I think we have to accept the fluidity of the pattern in that the pattern is constantly evolving, the gfs 06 hrs run IMO has the right idea but then doesn't follow through with the most likely scenario, the high should transfer towards Iceland and then a trough develop over Scandi, the troughing over the eastern USA in line with a very positive PNA should be aligned north south with WAA building pressure over Greenland, its a reload until the upstream pattern changes. We should then see another northerly before yet again the reload back to easterly.

I would be very surprised if the ensembles don't have quite a few members with those synoptics.

I agree Nick and your opening statement should be the foremost idea to bring to mind when analysing any model run.

"I think we have to accept the fluidity of the pattern in that the pattern is constantly evolving"

[cooling climate]

GP's in depth analysis this morning looks spot on the money to me and this little wobble from the

models, (more especially 06z GFS) should I think be taken as no more than that.

As with any pattern change IE the east/northeasterly retrograding to a block over the Iceland area

will probably take the models a day or two to sort out.

[bluearmy]

interestingly, TS on N24 commented that tuesdays front will give rain/sleet/snow as it comes through the midlands and further south. unlikely to be settling he said. the models show low enough thicknesses and uppers i would have thought for this not to be the case. there is a small warm sector os the front enters the north but surely this would be mixed out by the time it reaches the midlands.(and there was no issue with no snow north of the midlands). strange? (he did say that it woiuld be cold enough after this for snow beyond). NAE wont show enough info till late today to judge this.

[nick sussex]

when you say SHOULD Nick, do you mean scientifically as the programme says or or you more at hopecasting perhaps?

Of course you could be right and we both have to wait until that time to see whether you or the model is correct. Personally what the model is showing seems reasonable to me but then I tend not to follow each nuance of every run, especially at long time frames.

As to the ensembles-the whole idea of that is to step change the start data to see how many variations within a reasonable scale the model will then evolve. This then allows fairly complex statistics to be done an the likely % of any scenario being played out. All very complex and a bit beyond me I'm afraid. However, the ensemble idea when transferred to basic forecasting for a customer does allow degrees of probability for any particular forecast scenario to be given to the customer.

I don't do hopecasting especially as given some of the ridiculous criticism aimed at some of the regulars in the regional threads you have to be very careful what you say in here given the propensity for some people to throw toys out of their pram!

I'm basing my thoughts on the overall teleconnections and the expected upstream pattern, we'll see how it pans out but i think that seems a sensible way forward. As ever time will tell.

[microclimate]

Can I bank the 06z, please?? The first chart within the reliable(very 'ish') timeframe that gives ME potential of a good dumping. My first snowflake of the winter would be followed in very quick succession by many others if this verifies!

This has been hinted at for a while on the charts and seems a pluasable outcome. As the front drops down through us it is heading towards where the energy is, a formation of a small trough before it gets absorbed into the parent Med one, why not? Wishful thinking maybe but this is the first time this winter I have been able to see an evolution within 4 days that COULD give me down here a bit of snow.

But as we know the models giveth and the models taketh away.......

[weather eater]

when you say SHOULD Nick, do you mean scientifically as the programme says or or you more at hopecasting perhaps?

Of course you could be right and we both have to wait until that time to see whether you or the model is correct. Personally what the model is showing seems reasonable to me but then I tend not to follow each nuance of every run, especially at long time frames.

As to the ensembles-the whole idea of that is to step change the start data to see how many variations within a reasonable scale the model will then evolve. This then allows fairly complex statistics to be done an the likely % of any scenario being played out. All very complex and a bit beyond me I'm afraid. However, the ensemble idea when transferred to basic forecasting for a customer does allow degrees of probability for any particular forecast scenario to be given to the customer.

The ensembles are another one of those things I find odd in this model malarkey, far too often I see consecutive model runs, lets say the 00z and the 12z ECM operational runs, that are vastly different from each other, yet both are supported by their ensemble suite. Obviously their starting data is different. but it should not be so much as to give such widely different ensemble results. Often we then see the following days 00z back up the previous days 00z. Surely what should occur in that scenario, is that the sandwiched 12z operational is an outlier within its ensemble suite, but not backed by them. My own method is to ignore the ensembles and to keep all the days operational’s from the GFS and ECM in my minds eye and use those as an ensemble, then try and make a judgement on future evolutions from that.

[VillagePlank]

Morning, all.

Well, the North Sea looks drier than ever; RH down to 60% for parts of the week, but generally 70%. I take on board comments, last night, that the North Sea can herald surprises, so I must add the caveat that GFS can model this extremely badly, and experience counts for a lot in these scenarios.

However, CAPE, and LI levels are better, this morning: a feature moves south and westwards from SW Norway which has reasonble LI, CAPE, and VV@700hPa associated with it. It does look to weaken considerably before making landfall, but if the clouds are formed, precipitation should happen for areas like King's Lynn. This feature looks like a dew-point depression boost.

Precipitation amounts are much higher for all areas south of the M4 this morning mid-week and after, with the north/eastern coasts still in the firing line for the entire week. Perhaps 30cms accumulated by Friday for these areas.

Low pressure forming in channel gives a huge boost to central southern England, from this run, and is now the most favoured location before it exits south east which heralds high pressure taking over from the north west. However, this feature looks to come and go across members and models - so not sure on it's reliability. A curious front looking feature engages the far the south and east at the end of the week, which could boost levels there

Overall very cold, again.

Edited January 3, 2010 by VillagePlank

[johnholmes]

The ensembles are another one of those things I find odd in this model malarkey, far too often I see consecutive model runs, lets say the 00z and the 12z ECM operational runs, that are vastly different from each other, yet both are supported by their ensemble suite. Obviously their starting data is different. but it should not be so much as to give such widely different ensemble results. Often we then see the following days 00z back up the previous days 00z. Surely what should occur in that scenario, is that the sandwiched 12z operational is an outlier within its ensemble suite, but not backed by them. My own method is to ignore the ensembles and to keep all the days operational's from the GFS and ECM in my minds eye and use those as an ensemble, then try and make a judgement on future evolutions from that.

that may no be a bad idea

my mind could not cope with that so when I was doing the 12z GFS checks over a 3 year period I simply compared 12 to 12 each day.

It did seem to show a reasonable degree of conformity not always of course but more often than not. That is when, at times scales beyond T+144, sometimes as low as T+72, I decided not to favour comparing each run.

As to ensembles I'm never sure if the major centres took them up as much of their literature would have you believe-to provide a better service to their customers.

OR

because, and its much as I've posted for several years on here, the acceptance that no large scale improvements are going to occur in weather model performances under many decades if that. Trying to get the huge sums involved for computers means that the major centres need their or their area(EC) governments to stump up huge amounts of cash. By showing that ensemble forecasting can give a rather better service to some of their major customers then that could be the way?

just a thought.

ps

have a look at the NOAA model charts for their own and other models 500mb and other levels accuracy over the past 10 years!

Edited January 3, 2010 by johnholmes

[johnholmes]

I don't do hopecasting especially as given some of the ridiculous criticism aimed at some of the regulars in the regional threads you have to be very careful what you say in here given the propensity for some people to throw toys out of their pram!

I'm basing my thoughts on the overall teleconnections and the expected upstream pattern, we'll see how it pans out but i think that seems a sensible way forward. As ever time will tell.

Just which set of time scales are you referring to Nick please?

[Snowyowl9]

After the the end of this week this will make major headlines for snow for the eastern side of the country in particular,heaviest snowfall since 1991 maybe since 1987 there,but not as cold of course that was a much straighter easterly,this is much of a north with a slight easterly element,surface winds for the bulk of the week will be north.

Country tracks going for widespread 0c maxes thats 4c higher to January 1987.

Todays maxes over the snowfields.

http://91.121.93.17/wz/pics/Rmgfs124.gif

Edited January 3, 2010 by Snowyowl9

[weather eater]

that may no be a bad idea

my mind could not cope with that so when I was doing the 12z GFS checks over a 3 year period I simply compared 12 to 12 each day.

It did seem to show a reasonable degree of conformity not always of course but more often than not. That is when, at times scales beyond T+144, sometimes as low as T+72, I decided not to favour comparing each run.

As to ensembles I'm never sure if the major centres took them up as much of their literature would have you believe-to provide a better service to their customers.

OR

because, and its much as I've posted for several years on here, the acceptance that no large scale improvements are going to occur in weather model performances under many decades if that. Trying to get the huge sums involved for computers means that the major centres need their or their area(EC) governments to stump up huge amounts of cash. By showing that ensemble forecasting can give a rather better service to some of their major customers then that could be the way?

just a thought.

ps

have a look at the NOAA model charts for their own and other models 500mb and other levels accuracy over the past 10 years!

Thanks for that John; at least I'm not the only one who views them in a slightly suspect fashion.

I’m happy to stick to the method I use.

Keep recent model runs in mind and try to spot where the point of deviation is and not bother to much with what comes after, until such time a clear trend emerges.

Ignore precipitation charts beyond 24-36hrs.

Accept that even in a clear trend, small changes in detail make big differences on a small island.

Look but don’t take too seriously output beyond 144hr, sometimes less.

Don’t get caught up in ramping just because everybody else is.

Don’t dismiss output just because it doesn't show what you want to see.

Accept that even in a clear trend, model runs may randomly jump ship, this is not an indication of a new trend until it has support, and there is no need to reach for the Prozac.

Pay attention to those with proven knowledge in the field of metrology.

Beware of members who may have a big reputation but whose forecasts have all to often fallen over, even if they appear to be on the ball in present situation, ask yourself how often are they really on the ball.

Realise that climate is dynamic and constantly evolving, model runs never truly verify, even if its only small detail changes, we live on a small island, small can be big here.

[bluearmy]

despite ts enthusiasm for the snow streamer into the se at the end of the week, the GEFS are definitely trending towards a cross between the GEM and ECM/GFS scenarios whereby the high drifts se closer to us and then merges with the rise in pressure to the ne. either way, it looks as though the possible squeeze of easterlies later in the week which brings the biggest chance of widespread snowfall is currently being pushed away from our shores. we have often seen trends come and go but this current trend which also pushes the coldest uppers to our se would be par for the course for our part of nw europe. we rarely get a direct hit. (last feb being the first for a long time - prob 1995 before that).

[john w]

Upgrades (I think) to GFS for a snow event in the central south on the 00z ad 06z. This despite my comments last night that I felt the flow would be too slack and any precip would fizzle. The GFS at least intensifies the pressure. The latest ECM also doing the same which would allow for...as I said, stronger flows across the front, more uplift and hopefully (though personally snow gives me a headache at home and work) more snow.

I do have a question however and it is related to outputs and the meto output.

Without giving the game away several government authorities rely on bespoke meto forecasts to go about their duties. The ones we get have a good deal of linguistic interpretation and a summary of many models. They then have associated confidence bands for each model as well as verification analysis (much like the ones posted on this forum). I was wondering, however, if the METO on their main forecasts also undertake this task, or if they output forecasts based entirelyon their own model run. Also...do the BBC use data solely from the METO as opposed to any other models??

Thanks in advance for any responses!

[Frosty.]

This is what GFS 06z op run last Sunday was forecasting for this Sunday

And from a fortnight ago for this Sunday

Once again Mr_Data you have shown how utterly poor the gfs is in the T+168 range, I often read on here how the 6z is the garbage run, let's wait and see what the 12z runs bring.

[Bristawl Si]

Thanks for that John; at least I'm not the only one who views them in a slightly suspect fashion.

I'm happy to stick to the method I use.

Keep recent model runs in mind and try to spot where the point of deviation is and not bother to much with what comes after, until such time a clear trend emerges.

Ignore precipitation charts beyond 24-36hrs.

Accept that even in a clear trend, small changes in detail make big differences on a small island.

Look but don't take too seriously output beyond 144hr, sometimes less.

Don't get caught up in ramping just because everybody else is.

Don't dismiss output just because it doesn't show what you want to see.

Accept that even in a clear trend, model runs may randomly jump ship, this is not an indication of a new trend until it has support, and there is no need to reach for the Prozac.

Pay attention to those with proven knowledge in the field of metrology.

Beware of members who may have a big reputation but whose forecasts have all to often fallen over, even if they appear to be on the ball in present situation, ask yourself how often are they really on the ball.

Realise that climate is dynamic and constantly evolving, model runs never truly verify, even if its only small detail changes, we live on a small island, small can be big here.

I must admit i'm taking less note of the models past 48 hours these last couple of weeks. For those saying everything's set in stone and best synoptics for years, nothing's been delivered yet in the West. At the mo i see Scotland and N.E. england as having a pretty exceptional winter (compared to last 20 years).

It's not even that cold here in the West. Ok, colder than recent years, but defo not as cold as some of the winters experienced many many years ago.

From a snow viewpoint we've had one half decent snowfall, from a lucky group of showers spawned from the Bristol Channel. this aint no 79, 81 or 87 here in the West.

as for the model predictions for 2nd half of the week, things could change. I noticed some of the faxes for Tuesday showing slightly warmer air in the mix for western parts (well it's only 2 days ago that temps were forecast to be freezing during day even here in the West). As this island of ours is only a few hundred miles wide, it just needs a small change in actual positions of different air masses to change things from both a regional and national perspective.

to those newer members on here, i'd advise to wait and see what comes - believe me, model runs do not dictate this island's weather.

[bluearmy]

although the ecm looked less cold, under a col with snow cover, look at the temps for london

http://www.meteogroup.co.uk/uk/home/weather/services/ensemble_forecast.html

[MR EXTREMES]

i see the mood has gone downbeat from 1 run this morning still looks very exciting and if anyone thinks its time to throw in the towel then think again.

might be worth taking a look at gp post this morning,

i have had no snow but some really nice cold frosty mornings,

i know the snows will come in any great snowy winter cold sets in first,

this is whats happening.

the models are suggesting this.

and id say forget this mornings run until all models come into agreement im sure things will be brighter later today they always are

Edited January 3, 2010 by badboy657

[Snowyowl9]

despite ts enthusiasm for the snow streamer into the se at the end of the week, the GEFS are definitely trending towards a cross between the GEM and ECM/GFS scenarios whereby the high drifts se closer to us and then merges with the rise in pressure to the ne. either way, it looks as though the possible squeeze of easterlies later in the week which brings the biggest chance of widespread snowfall is currently being pushed away from our shores. we have often seen trends come and go but this current trend which also pushes the coldest uppers to our se would be par for the course for our part of nw europe. we rarely get a direct hit. (last feb being the first for a long time - prob 1995 before that).

Last february was very good indeed even here for frequent snow showers during that coldest day,wind was actually N/NE.

Better hope GFS 0z run comes off and UKMO,as ECM shows very light winds and GFS 6z just ends up very cold but mainly dry,strong winds I want.

http://91.121.93.17/pics/Rukm1441.gif

Edited January 3, 2010 by Snowyowl9