General Model Output Discussion

[cold snap]

Could there be disagrement right up to the event.??

Put GFS and UKMO in a ring and let them sort it out.those two models hate each other right now

[Harsh Climate]

I have to disagree here, you have to take a models bias into consideration, notice that feature that the gfs had near Norway in its earlier output which the ecm filled quickly, yes another case of the gfs overblowing these features,the ecm in all the time i've been viewing it has only had to do one major backtrack within the 96hrs timeframe, the gfs numerous, i'm sorry but the verification stats say it all, the gfs is in 4rth place behind the ecm, ukmo,and gem. The gfs even allowing for its early disagreement has caved in to the ecm regarding the undercutting low later because yes you guessed it, it modelled the feature near Norway incorrectly by overblowing it!

I could understand this obsession by many members in following the gfs if it was the best global model but it isn't, all models do go through the odd bad patch but the gfs has many more of these than the ecm.

I see what you mean. But I guess you could pick little features/differences that one model has backtracked on towards a different model (gfs to ecm for example), for the ecm then to backtrack towards the gfs on a another feature...

I think if we forget the bigger picture and look at how the gfs has performed over ems for our weather within 72-96h, I think the GFS is never too far off the mark and any differences in that range seem to be gradual run to run..

I think the ecm is great at picking up new trends and signals at a longer range but I tend to find that model change quite drastically run to run, you never know which to believe lol.. Thats why I would allways look for backing from ukmo and gfs before taking the ecm as most likely option

EDIT: I must admit though, the GFS is showing exactly what I want to see right now, so I may just be a teensy weensy bit biased at the moment

Edited December 28, 2009 by Jed Bickerdike

[Paul Sherman]

Lots of Talk of how the Different Models perform on this thread etc

Have stated this before but GFS Is much better at the Mid to Long Term with the ECMWF & UKMET Much better at closer range of T24 - T72

Take last Mondays Snow debacle along the Hampshire area for Instance.

UKMET & ECM Had the Precip further North and even the Meto had the front pivoting and then the Cold Front moving East much later in the day, this was pretty much spot on almost to the 10-50 Mile range and they nailed that at T48.

On the same situation last monday GFS Did not even have the Precip advancing further North than Sussex and Kent and kept 95% of the precip over in NW France.

Just take note of how the Models perform, like I said last week when chasing Tornadoes we use 3 different Models as do most of the Chasers

GFS For the T168 - T240

ECM For the T36 - T168

RUC For the T0 - T24

Paul S

[Geordiesnow]

I have to admit, i thought the GFS would of come in line by now but instead, we now got the UKMO now starting to have the low too far North. Although the FAX's has the low further South, i still think with the UKMO coming on board is a concern.

GFS in the last 6-8 runs have shown the low to be further North than we all liked, some pretty good consistancy it must be said. I bet those who were doubting the GFS are now a little worried about the GFS being right and the popular ECM/GEM actually getting the wrong.

I won't be too surprised if we signs of the ECM coming on board tonight, but i hope i am wrong but i can't really see it backtracking now, surely it would of done that earlier??

[MR EXTREMES]

Btw for some reason I've just been reminded of this chart:

http://www.wetterzen...00120090210.gif

This was the only other time I remember such disagreement at close range - the ECM had the track of that low further N than actually occurred and the GFS turned out to be (almost) right in all its insistence. In fact the ECM ended up being by far the weakest model and I feel we may be making the error of relying on it too much at short range (it is not designed for this purpose).

The GFS/UKMO may not be as wrong as most of us would like. Personally I'm dreading the GFS being right as the weather would be truly revolting here!

well i never herd of 4 models all gettting it wrong.

as far as i can see it gem jma ecm ukmo against the gfs:whistling:

id be very dissapointed if all them failed but not likely though.

[nick sussex]

I see what you mean. But I guess you could pick little features/differences that one model has backtracked on towards a different model (gfs to ecm for example), for the ecm then to backtrack towards the gfs on a another feature...

I think if we forget the bigger picture and look at how the gfs has performed over ems for our weather within 72-96h, I think the GFS is never too far off the mark and any differences in that range seem to be gradual run to run..

I think the ecm is great at picking up new trends and signals at a longer range but I tend to find that model change quite drastically run to run, you never know which to believe lol.. Thats why I would allways look for backing from ukmo and gfs before taking the ecm as most likely option

I agree that we need to see good model agreement however lets put FI aside and in dealing with the gfs versus ecm within 144hrs the ecm is superior to the gfs. The gfs is more likely to be way out within this timeframe than the ecm, the ecm when it goes wrong doesn't do it so spectacularly!

Here are the verification stats for the ecm just against all 4 gfs runs upto 120hrs.

http://www.emc.ncep.noaa.gov/gmb/STATS/html4x/acz5.html

The ecm beats every single gfs run in both the northern and southern hemispheres, in terms of FI IMO both models can go completely AWOL but in this case we don't have access to any verification stats, what we do have shows without any doubt the ecm is the best global model upto 144 hrs.

Edited December 28, 2009 by nick sussex

[Harsh Climate]

Lots of Talk of how the Different Models perform on this thread etc

Have stated this before but GFS Is much better at the Mid to Long Term with the ECMWF & UKMET Much better at closer range of T24 - T72

Take last Mondays Snow debacle along the Hampshire area for Instance.

UKMET & ECM Had the Precip further North and even the Meto had the front pivoting and then the Cold Front moving East much later in the day, this was pretty much spot on almost to the 10-50 Mile range and they nailed that at T48.

On the same situation last monday GFS Did not even have the Precip advancing further North than Sussex and Kent and kept 95% of the precip over in NW France.

Just take note of how the Models perform, like I said last week when chasing Tornadoes we use 3 different Models as do most of the Chasers

GFS For the T168 - T240

ECM For the T36 - T168

RUC For the T0 - T24

Paul S

Well if thats how it is, thats tought me something new

[Yeti]

On the other hand Yeti the GFS did a poor job with the Feb 07 set-up and thats IMO the closest set-up we've had to this in the internet age with regards to the evolution of the low pressure system and the troughing of the low.

I remember the GFS kept driving the precip into N.England/Scotland whilst the UKMO/ECM kept holding the fronts further south and in truth the GFS proved to be vastly too progressive with the frontal system pushing northwards,in fact I think it was only after the second front finally came through did the first front reach further north a full 24-30hrs later then was progged...I suspect something similar will occur here.

Interesting, thanks...

The UKMO 6z is certainly a lot better than the 0z and is MUCH more similar to the ECM which is encouraging!

I know people are beginning to complain about the endless chatter about the track of the low but that's simply because it's two days away and will mean the difference between perhaps a foot of snow and absolutely nothing for some places. Despite the medium term prospects looking excellent, I imagine most people on here would rather grab a foot of imminent snow than wait for the next pulse of cold air that may or may not have similar snow potential for them!

[Altostratus]

I see what you mean. But I guess you could pick little features/differences that one model has backtracked on towards a different model (gfs to ecm for example), for the ecm then to backtrack towards the gfs on a another feature...

I think if we forget the bigger picture and look at how the gfs has performed over ems for our weather within 72-96h, I think the GFS is never too far off the mark and any differences in that range seem to be gradual run to run..

I think the ecm is great at picking up new trends and signals at a longer range but I tend to find that model change quite drastically run to run, you never know which to believe lol.. Thats why I would allways look for backing from ukmo and gfs before taking the ecm as most likely option

EDIT: I must admit though, the GFS is showing exactly what I want to see right now, so I may just be a teensy weensy bit biased at the moment

I agree with Nick on this one. For the UK, the ECM is usually more accurate. The verification data proves this. I believe the ECM is the model with the highest resolution too. The ECM also has a large ensemble suite that has been very bullish for the past few days. The GFS ensembles have been up and down like a yo-yo. The ECM is seen by many profesional organsiations as the best model. I agree tha the GFS data cannot be discounted, but in a head-to-head contest with the ECM, the ECM is rarely wrong in these situations.

[AtlanticFlamethrower]

Model disagreement in pictures.

NOGAPS 00z has the trough over Brittany

ECM 00z has the trough over Brittany

GEM 00z has the trough in Alsace-Lorraine

GME 00z puts the trough just north of Paris (chart 12 hours earlier than the others)

UKMET 00z puts the trough in the channel

GFS 06z places the trough over Kent

UKMET/GFS versus the Rest. the UKMET has a more southerly low than the GFS. the GFS is about 24 hrs more progressive than the UKMET.

[Paul Sherman]

Well if thats how it is, thats tought me something new

Jed

See the Verification Stats that Nick S has just posted up If you want further proof.

BTW I think your area is going to get hammered anyway as I see this being a bit in between GFS & UKMO This Time.

The problem areas are places within 10-40 Miles of the East Coast as the North Sea is like a Tepid Bath and with those Easterlies and no cold pool over the Continent we are scr**d more esp the further South and East you are.

I just cannot see anywhere in a line East of Herts up to W Norfolk getting anything until the colder air digs south on New Years Eve/Night to New Years Day.

Paul S

[Geordiesnow]

I agree that we need to see good model agreement however lets put FI aside and in dealing with the gfs versus ecm within 144hrs the ecm is superior to the gfs. The gfs is more likely to be way out within this timeframe than the ecm, the ecm when it goes wrong doesn't do it so spectacularly!

Here are the verification stats for the ecm just against all 4 gfs runs upto 120hrs.

http://www.emc.ncep.noaa.gov/gmb/STATS/html4x/acz5.html

The ecm beats every single gfs run in both the northern and southern hemispheres, in terms of FI IMO both models can go completely AWOL but in this case we don't have access to any verification stats, what we do have shows without any doubt the ecm is the best global model upto 144 hrs.

but stats means ZILCH, just because statistics would suggest the ECM is a better global model than the GFS does not mean the GFS is wrong this time and the ecm is right.

Just because recent performances have been good for the ECM does not mean, it is totally reliable.

For all we know, the GFS could be onto something here and the ECM could easily backtrack tonight. I think some are over complacent regarding recent performances and model bias and actually forgetting that the GFS has been consistent regarding this low for a good old 6-8 runs now. In fairness so has the ECM whilst the UKMO has been in and out of both scenario's.

I can't say without much confidence which one would be right but you just know our luck, the GFS is probably going to be closer to the mark. Lets just hope it backtracks on its 12Z output and at the very latest the 18Z output.

[Duane S.]

I have to admit, i thought the GFS would of come in line by now but instead, we now got the UKMO now starting to have the low too far North. Although the FAX's has the low further South, i still think with the UKMO coming on board is a concern.

GFS in the last 6-8 runs have shown the low to be further North than we all liked, some pretty good consistancy it must be said. I bet those who were doubting the GFS are now a little worried about the GFS being right and the popular ECM/GEM actually getting the wrong.

I won't be too surprised if we signs of the ECM coming on board tonight, but i hope i am wrong but i can't really see it backtracking now, surely it would of done that earlier??

Me too regards the GFS coming into line. Although I suppose the METO and the GFS have edged towards each other a tad, BUT the all important fax outputs don't follow either model, favouring something closer to the ECM, so I am very encouraged by that.

I have never been in the firing line for this initial 'event' so no bias from me but I think we are going to see more of a GFS backtrack than an ECM one come the event. The GFS is simply not as good short range as most of the others. Mid term is where it is normally best.

That said, it is far from impossible that is has got it right. I remember one occasion (last February I think?) when the GFS was out on its own regarding the track of a SW over us and ended up being smack on, so it is certainly no done deal yet...

Edited December 28, 2009 by s4lancia

[Altostratus]

but stats means ZILCH, just because statistics would suggest the ECM is a better global model than the GFS does not mean the GFS is wrong this time and the ecm is right.

Just because recent performances have been good for the ECM does not mean, it is totally reliable.

For all we know, the GFS could be onto something here and the ECM could easily backtrack tonight. I think some are over complacent regarding recent performances and model bias and actually forgetting that the GFS has been consistent regarding this low for a good old 6-8 runs now. In fairness so has the ECM whilst the UKMO has been in and out of both scenario's.

I can't say without much confidence which one would be right but you just know our luck, the GFS is probably going to be closer to the mark. Lets just hope it backtracks on its 12Z output and at the very latest the 18Z output.

You are absolutely right. For all we know, the GFS might well have called this one right. You win some, you lose some. Based on the statistical analysis, the odds favour the ECM outcome however. It is a difficult one to call and I’m sure the professionals are debating it as much as us.

[Geordiesnow]

Me too regards the GFS coming into line. Although I suppose the METO and the GFS have edged towards each other a tad, BUT the all important fax outputs don't follow either model, favouring something closer to the ECM, so I am very encouraged by that.

I have never been in the firing line for this initial 'event' so no bias from me but I think we are going to see more of a GFS backtrack than an ECM one come the event. The GFS is simply not as good short range as most of the others. Mid term is where it is normally best.

That said, it is far from impossible that is has got it right. I remember one occasion (last February I think?) when the GFS was out on its own regarding the track of a SW over us and ended up being smack on, so it is certainly no done deal yet...

At least the good news is that the UKMO 06Z has apparently has the low further South and hopefully this is a good sign of backtracking from the 00Z output.

I must admit, if your a neutral and you are pretty much guaranteed snow from either set up or you are not at all fussed what is going to happen, then the models are very interesting. If not, fingers nails being bitten to the bone type of nerves all the way.

[Harsh Climate]

Jed

See the Verification Stats that Nick S has just posted up If you want further proof.

BTW I think your area is going to get hammered anyway as I see this being a bit in between GFS & UKMO This Time.

The problem areas are places within 10-40 Miles of the East Coast as the North Sea is like a Tepid Bath and with those Easterlies and no cold pool over the Continent we are scr**d more esp the further South and East you are.

I just cannot see anywhere in a line East of Herts up to W Norfolk getting anything until the colder air digs south on New Years Eve/Night to New Years Day.

Paul S

Thanks for that

Ye that clearly does prove that the ECM is the most accurate model

[snowking]

Jed

See the Verification Stats that Nick S has just posted up If you want further proof.

BTW I think your area is going to get hammered anyway as I see this being a bit in between GFS & UKMO This Time.

The problem areas are places within 10-40 Miles of the East Coast as the North Sea is like a Tepid Bath and with those Easterlies and no cold pool over the Continent we are scr**d more esp the further South and East you are.

I just cannot see anywhere in a line East of Herts up to W Norfolk getting anything until the colder air digs south on New Years Eve/Night to New Years Day.

Paul S

Very much agreed - A1(M) looks to be the eastern-most flank for snow unless we see a dramatic shift SE'wards of the LP for now

Even when the colder air digs in over the new year i'd imagine a lot of the precipitation would have died out before it reaches us again here, so those hoping for something in the SE might have to hope for an extra trough/frontal development

SK

[Snowaddict]

In terms of conifdence, more cause for optimism..

The latest HIRLAM (which is a HI-Res Model used by the Irish Met) - very similar to that of NAE from UKMO also agrees with the NAE 06Z Hrs output..

NAE suggests snowfall for a large stretch of southern Ireland later tomorrow and this is reflected in the latest BBC Precip charts which show extensive snowfall across Wales, Southern Ireland and parts of the midlands..

In fact, the NAE 06 Hrs output is almost fantastic..

The HIRLAM Hi-Res Model also keeps the low to the south as with the NAE..

It also develops a small secondary low just off Cornwall towards T+48 as on the overnight ECMWF output..

Latest 5 Day stats are out and guess which model leads the pack - ECMWF..

http://www.emc.ncep.noaa.gov/gmb/STATS/html/graphics/acz5.gif

SA

[cooling climate]

It is a shame really how the GFS is struggling with this upcoming freeze when it performed so well

in the last spell of wintry weather (apart from Scotland where it has seen one continuous cold spell).

The 0z GFS run showed the wintry weather lasting throughout the run and then along comes the 06z

and again it sends the block in completely the wrong direction and we end up in a south/southeasterly

airflow.

I still can not understand why people are giving so much attention to the GFS when it is obviously

struggling with this particular set up. Just because it handled the last cold spell better or at least led the

way does not mean it will do so every time which is becoming quite clear this time around.

The low off the Danish coast which was modeled yesterday to anchor there for two to three days holding up the

easterly airflow has now vanished without trace and that alone should have told us that it will be overdoing

shortwaves etc and should not be trusted.

Stick with the faxes for the short term and the ECM and even the GEM further afield as these two models have

not wavered in their output.

A long wintry freeze is on the way for UK and much of Europe.

[nick sussex]

but stats means ZILCH, just because statistics would suggest the ECM is a better global model than the GFS does not mean the GFS is wrong this time and the ecm is right.

Just because recent performances have been good for the ECM does not mean, it is totally reliable.

For all we know, the GFS could be onto something here and the ECM could easily backtrack tonight. I think some are over complacent regarding recent performances and model bias and actually forgetting that the GFS has been consistent regarding this low for a good old 6-8 runs now. In fairness so has the ECM whilst the UKMO has been in and out of both scenario's.

I can't say without much confidence which one would be right but you just know our luck, the GFS is probably going to be closer to the mark. Lets just hope it backtracks on its 12Z output and at the very latest the 18Z output.

I'm not disagreeing with you regarding the fact that all models can go wrong even the ecm, but we need something to judge them by, it might not be perfect but the verification stats are all we have. I was just pointing out that generally across the last few years the ecm has been most reliable, i never said it was perfect. This will be my last word on this subject until the next person goads me into a response!

These are some of the observed biases of the gfs by NOAA, do any of these ring a bell with members in here?

Prediction of southward progression of cold air over done

Model a bit too extreme in temp patterns beyond 84 hours

Precip Type Algorithm off of GFS too eager to depict snow

Over forecast of cyclogenesis east of Rockies

These were observed across the USA but would still apply upstream. Thats not to say the ecm doesn't also have some biases one of which seems to be in recent years overblowing northern blocking in its FI, but NOAA haven't actually reported anything on their site.

Anyway i think we should move on as i feel like i'm flogging a dead horse here! We'll know by this evening which model has called the early low track the best,hopefully!!!

[HotCuppa]

and I think its pretty ridiculous to say stats matter..because they are meaningless in the grand scheme of things. Like how the ECM messed up at the start of this cold spell ?? I am no fan of the GFS but there is too much wishing for snow here. Perhaps the new year for places north of N wales

its not who is the best, but what is going to happen with this low in the coming few days. LP to far north. Warnings have now been reduced to possible advisories for north west england infact its quite a big back track there by the meto office.

Edited December 28, 2009 by HotCuppa

[Tucco]

12z Meto Invent up date confirms SWW, heaviest precipiation across Wales. Precipiation moves Northwards and Westwards weakening across N England. Heavy precipiation bands contuine move in S Wales/S Mildands at 2100z Tuesday.

Mark

Teesdale

[nick sussex]

and I think its pretty ridiculous to say stats matter..because they are meaningless in the grand scheme of things. Like how the ECM messed up at the start of this cold spell ?? I am no fan of the GFS but there is too much wishing for snow here. Perhaps the new year for places north of N wales

Is there a chair I can talk to? have you read any of the posts made by more coherent people? Stats matter across all walks of life as thats basically the only measure that people can't argue over as long as their properly verified! i've put the stats up countless times regarding the models. There is no other way of judging the models that we have access to, all that would happen is a load of spurious arguments not based on fact. Now for the last time the ECM is the most reliable global weather model, this is a FACT!

[HotCuppa]

and to me its also a fact they are wrong now, like they where wrong at the start of the cold spell but time will tell i suppose with this. wow you are confrontational. We are talking about what's going to happen with this current setup not what the stats are. I have never once said that the ECM wasnt the bee's knees. All I care about is where this low will end up and whether we get either snow or rain. As for the stats they can change over time..just because they are the global model to use now, doesn't make them absolutely right absolutely every time but thats not what you where saying anyways as for coherent..personally I think that was a bit uncalled for wasn't it?

Edited December 28, 2009 by HotCuppa

[Guest North Sea Snow Convection]

IMBYISM is behind a lot of the current model preference and 'will it snow' debate. As someone else said earlier, as one who has always believed I will see nothing from this marginal low pressure snow event, I am quite ambivalent about which model is 'right' in terms of the short term this week. In terms of favouring the ECM and the GEM, I can't (and won't) add anything to the sensible posts of people like climate cooling, nick sussex, snow addict and a few others - but the main interest for guidance to me is following these models in terms of the evolution of the pattern/cold spell 'post low pressure'. The reason for this is that they have been (so far) much more reliable, less erratic and more consistent

We are getting to the stage now, if I am to be totally honest, where a lot of the regional IMBY stuff about this low should, imo, be going into the regional threads. Put it this way, if I was in the probable snow zone, that is where I would be now be posting about this event.

Edited December 28, 2009 by North Sea Snow Convection