Model Output Discussion 13/11/13 >

[Weathizard]

Such a huge difference even at t96, only 4 days away which is normally within the time limit for when things don't change much, one of the big 2 is going to have egg on the face big time.

[Tom Jarvis]

Gfs could give a northerly soon 

[SMU]

Isn't GFS supposed to be 'cannon fodder', especially the 06?

 

The GFS is cannon fodder, however eventually it will catch up....

It wont catch up this run but it will edge closer- I guarantee you at 192 it will still have a UK high-

 

A bit like when people don't read my posts in detail - eventually they catch up......

 

 

S

Edited November 18, 2013 by Steve Murr

[Matty M]

The GFS is cannon fodder, however eventually it will catch up....

It wont catch up this run but it will edge closer- I guarantee you at 192 it will still have a UK high-

 

A bit like when people don't read my posts in detail - eventually they catch up......

 

 

S

Pretty silly to call the GFS model a cannon fodder model IMO. The GFS is only marginally behind the UKM/ECMWF model and is by far the 3rd most accurate medium-term model.

 

It offers us so much data free to use and has enabled most on here a depth of knowledge in different model parameters that they would never of had without it.

 

It has also called a number of these situations right in the past.

Edited November 18, 2013 by Matty M

[Yarmy]

Such a huge difference even at t96, only 4 days away which is normally within the time limit for when things don't change much, one of the big 2 is going to have egg on the face big time.

Perhaps they both will. The GFS 06z is completely different to its 00z run even at T120

 

 

 

If it can change that much in 6 hours, it's going to be a long old week.

[Frosty.]

The 6z has very good verification stats for a cannon fodder model I have been told. :-)

[bluearmy]

perhaps time to remind ourselves that ecm op is a little biased towards excess amplification in its latter stages - at least the old version of the op was. guess this is the first test if the same is true of the new one !

Edited November 18, 2013 by bluearmy

[A.J]

Do I detect a hint of point scoring creeping into the thread again?.......A couple of members seem to be spoiling for a fight, so lets keep it clean folks.....on-topic discussion and debate is cool, but repeated below the belt blows will result in a definite 'points reduction' as such

[SMU]

Pretty silly to call the GFS model a cannon fodder model IMO. The GFS is only marginally behind the UKM/ECMWF model and is by far the 3rd most accurate medium-term model.

 

It offers us so much data free to use and has enabled most on here a depth of knowledge in different model parameters that they would never of had without it.

 

It has also called a number of these situations right in the past.

The 06z is - it always will be.

 

Its verification stats for Europe/ atlantic & north pole was probably nothing better that the NOGAPS was-

 

If you post up the 00z 102 & the 06z 96 it sort of proves my point

 

Im not saying the ECM is totally correct, however there is only 1 model going through rapid change & that's the GFS.

[MPG]

Pretty silly to call the GFS model a cannon fodder model IMO. The GFS is only marginally behind the UKM/ECMWF model and is by far the 3rd most accurate medium-term model.

 

It offers us so much data free to use and has enabled most on here a depth of knowledge in different model parameters that they would never of had without it.

 

It has also called a number of these situations right in the past.

Im no expert, but the differences between the 0z and 6z are amazing, so maybe the proof is in the pudding?

[chionomaniac]

I think that we all know that the GFS isn't cannon fodder, but we also know Steve's thoughts on how it doesn't handle split jet flow when faced with a potential blocking scenario, so let's just leave it there.

[NickR]

Im no expert, but the differences between the 0z and 6z are amazing, so maybe the proof is in the pudding?

Just to say that, if we're trying to work out how models perform, comparing 2 successive runs on one day is proof of absolutely nothing. We need to be more scientific and circumspect than that. :-)

[swfc]

So the 0z and 6z show a big difference in i suppose a reliable time range but is this just down to the confusion in the movement of the atlantic low and not the models accuracy?ive know idea tbh but i find it hard to believe a model runs accuracy is governed by the time it runs!Lack off data ete? Anyway its what it is and altho it does lean towards the euros early on it does flatten out so one step forward as they say.Pv looks more confused than me atm so i guess over the next couple of days many various outcomes will come and go before the final solution on this current saga concludes.Interesting as ever altho the "rutting"amongst a few members is tiresome tbh and quite boring My gut feeling still is a middle ground with the euros but hey what do i know

[Jonan92]

Navgem keeps supporting ECM

 

[Frosty.]

The Gfs 06z op run gives the impression of being dragged through a hedge backwards whilst kicking and screaming towards the Ecm 00z..it's painful to watch..I thought it would have been colder than that in the south to be honest.

[MPG]

Just to say that, if we're trying to work out how models perform, comparing 2 successive runs on one day is proof of absolutely nothing. We need to be more scientific and circumspect than that. :-)

Scientific and circumspect? As I said the proof is there infront of you. You have answered your question, 2 runs 6 hours appart and massive differences in that time frame.

[NickR]

Scientific and circumspect? As I said the proof is there infront of you. You have answered your question, 2 runs 6 hours appart and massive differences in that time frame.

 

Any analysis of the value or reliability of anything - models or otherwise - cannot be based upon or said to be "proved" by one individual example. Come on, this is hardly controversial! I'm not saying the GFS is NOT flawed.. I'm just stating the fairly basic idea that saying "look, 2 runs, very different. QED" is meaningless. There is a reason that reliability is based on long-running verification statistics and not picking 2 particular successive runs alone.

[johnholmes]

Re the disussion about models aand in particular GFS and its accuracy levels from run to run. The only way to check that is the already available NOAA assessments for each run for each model as to how the 500mb charts show up. Yes, some folk argue against them, not effective for the European area, and various other reasons. However, only rigorous scientific checks such as they are will show the modellers just where the problems are. I think we can take it as read that NOAA are, as other model centres will be, constantly working on how to improve their model and trying to work out why, if they are, one model is more accurate or one run is more accurate than another.

It is a touch tedious to keep seeing this crop up, not so much showing it, but the slightly acid remarks at one another. No need for that.

[trisnow]

Any analysis of the value or reliability of anything - models or otherwise - cannot be based upon or said to be "proved" by one individual example. Come on, this is hardly controversial! I'm not saying the GFS is NOT flawed.. I'm just stating the fairly basic idea that saying "look, 2 runs, very different. QED" is meaningless. There is a reason that reliability is based on long-running verification statistics and not picking 2 particular successive runs alone.

Where are the verification stats? Does anyone have a link please? I thought I read earlier that the 06 was one of the worst performers?

[johnholmes]

Where are the verification stats? Does anyone have a link please? I thought I read earlier that the 06 was one of the worst performers?

 

go to the NOAA site, I'll get the link in a tic

 

this is one link

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

Edited November 18, 2013 by johnholmes

[Frosty.]

On the plus side, the Gfs 06z op run is rather cold throughout with night frosts, nothing mild at all which is something at least but I think the Ecm is the model to follow in these potentially cold and blocked patterns.

[NickR]

Where are the verification stats? Does anyone have a link please? I thought I read earlier that the 06 was one of the worst performers?

http://www.emc.ncep.noaa.gov/gmb/STATS_vsdb/

 

These are comparingt he 0z runs of each. You can change it to the 12z... I don't have the link comparing different GFS runs.

 

Overall ECM leads, with UKMO and GFS about level.

 

At 10 days for the Northern Hemisphere, GFS is actually currently leading ECM on the 0z runs, though at that range statistics are not great for either, of course.

Edited November 18, 2013 by NickR

[radiohead]

Where are the verification stats? Does anyone have a link please? I thought I read earlier that the 06 was one of the worst performers?

http://models.weatherbell.com/verf/gfs_nh_f120.pngScores are on the right hand side. 06Z as good or bad as the others, and often scores higher than the others.

[NickR]

The 4 GFS cycles compared:

 

http://www.emc.ncep.noaa.gov/gmb/STATS_vsdb/gfs4cyc/

 

Not much difference TBH at 5days. 00 and 12 slightly better.

[Captain Shortwave]

I feel that I need more convincing on the blocked pattern shown by the ECM, mainly because we were at this juncture last week and well it didn't turn out as well as many of us hoped. I guess the one positive is that we seem to be able to recycle the Atlantic ridge once it collapses towards the UK which means we do get further injections of cold air from the north or north east which gives some opportunity for some snow, and of course helps to cool the European landmass.