Winter Model Output Discussion 01/02/2013 12z onwards

[snowmadchrisuk]

Regarding next week , i'm not to sure GFS has it's temps rather high 6-7 deg on Wednesday . With 528 dam line covering the country , uppers between -5 and -10 .. Seems unlikely unless there expecting some strong Sunshine,

[Recretos]

I know this is the "classic" model discussion thread, but anyway, here are some additional weekly forecasts to help you with the trends, directly from Asia. They will probably get lost in the massive crowd of posts anyway.

JMA weeklies, latest run 31/1. For easier orientation: Atlantic is on top, under the ridge on the week 1.

T2m anomaly.

And the Korean (KMA) weekly (10-day) ensembles. They have proved to be quite good in the past.

Best Regards.

Edited February 2, 2013 by Recretos

[Methuselah]

Probably because it showed a breakdown of the last freeze 72 hours before it eventually did end, 3 whole days error and the ecm and ukmo had been going for the extended cold throughout, the gfs had to back down. As steve said last night, the gfs has an eastward bias and it handles cold blocking over the uk and europe poorly. The Ecm 00z is a fantastic run for prolonged very cold snowy prospects and would have similarities to feb 1986 which was a very cold blocked month.

But there have been times when the GFS's 'Atlantic bias' has proved correct. And December's 'Beast From the East' fooled them all - apart from RJS and Piers Corbyn - I think?

[Frosty.]

But there have been times when the GFS's 'Atlantic bias' has proved correct. And December's 'Beast From the East' fooled them all - apart from RJS and Piers Corbyn - I think?

The law of averages suggests the gfs will get it right sometimes, actually it has more chances of getting it right with 4 runs a day but when there is cold blocking for uk/europe it seems to be lagging behind the ecm/ukmo in a poor 3rd or worse.

[Jason M]

GFS says no again.

To be fair I'm not too bothered about the opp run after 192 hours. I do want to see a good trend in the ensemble suite though and it would be good to see a better control run.

Jason

[bluearmy]

the 12z will be interesting for two reasons

1) will ecm op churn out a fourth successive run with good fi continuity? we have seen three on the bounce before from the model followed by a flip back but i dont recall ever seeing four and then a flip

2) will gfs op continue its slow backtrack as evident on the 00z run (not so much on the 06z)

[mulzy]

Lots of differences between the ECM and GFS this morning, but probably the most key is how the low in the around the eastern seaboard is handled at day 6. THe ECM has a slightly deeper feature that is slightly more north west when compared to the GFS. Both models deepen that low significantly but the track seems to be slightly different, with the ECM taking that low more NNE while the GFS takes it NE.

Edited February 2, 2013 by mulzy

[snowking]

I'm not buying an Easterly. Reasons

1) Ec op a relative outlier vs its ensembles.

2) UKMO looks ok but not on board.

3) GFS op and control not buying in one bit.

4) GEM not interested

5) ECM to bullish with easterlies all winter.

6) GFS modelled last breakdown correctly (albeit delayed) whilst ECM was showing Day after Tomorrow stuff

7) GFS got yesterday's complex system right first.

GFS has modelled Tuesday's system beautifully - super consistent for about 10 days.

Hi Stu,

Whilst I think you are wise to keep a level head about things, I don't think many of the points you make above are really relevant here.

The first thing to bear in mind is point 6. The GFS modelled the breakdown of the last cold spell so many times that eventually it had to be correct....but its important not to ignore the fact it got it wrong so many times before! It perhaps shows the tendency we have always known that the GFS likes to revert to type. We should not assume, however, that this reverting to type will always be incorrect.

Points 5-8....you have to bear in mind that, in a strict sense, the weather from a number of weeks ago, and the associated model rights/wrongs, have very little bearing on our current potential cold spell and how long it may or may not last for. No amount of modelling will have any bearing on what the actual outcome will be, and whilst I take the point that the GFS may have modelled tuesday correctly (though I havn't verified that personally yet by having a look through the recent archives), I would argue that the UKMO was the first to be correct with yesterdays system as it never really had much interest in a particularly deep LP at any stage.

Kind Regards

SK

[Long haul to mild]

I'm not buying an Easterly. Reasons

1) Ec op a relative outlier vs its ensembles.

2) UKMO looks ok but not on board.

3) GFS op and control not buying in one bit.

4) GEM not interested

5) ECM to bullish with easterlies all winter.

6) GFS modelled last breakdown correctly (albeit delayed) whilst ECM was showing Day after Tomorrow stuff

7) GFS got yesterday's complex system right first.

GFS has modelled Tuesday's system beautifully - super consistent for about 10 days.

Have to disagree on point 7; the GFS and ECM both overblew yesterday's Low in the run up to it; however the UKMO never had such a strong feature and was proved correct.

And as for the last cold spell breakdown, the GFS was very wrong for a lot of runs before it finally got it right, which led to a number of snow events it had not forecast.

However, I will agree that ECM has been wrong about Easterlies a few times already this winter. For me, UKMO is the one to follow at present.

[TEITS]

I'm not buying an Easterly. Reasons

1) Ec op a relative outlier vs its ensembles.

2) UKMO looks ok but not on board.

3) GFS op and control not buying in one bit.

4) GEM not interested

5) ECM to bullish with easterlies all winter.

6) GFS modelled last breakdown correctly (albeit delayed) whilst ECM was showing Day after Tomorrow stuff

7) GFS got yesterday's complex system right first.

GFS has modelled Tuesday's system beautifully - super consistent for about 10 days.

1) Have you seen all the ECM ensembles or are you basing this on the mean?

2) UKMO is on board and would follow the ECM.

3) GFS/GEFS control have been crap all winter.

4) What do those canadians know!

5) ECM has only shown an E,ly in Dec and thats it.

6) So what? ECM did not show a day after tomorrow scenario, just prolonged the cold spell.

7) Are you kidding? Only the UKMO got yesterdays LP right, the GFS at times suggested a hurricane.

[bluearmy]

despite the 06z low res gfs looking uninspiring, it should be noted that almost the whole of europe to our east remains sub zero throughout. so the atlantic isnt getting into europe in any meaningful sense. the russian block responsible and there will be a battleground area between this cold air and the atlantic.

oh and dave, whilst i agree with what you posted above, i wonder re the gem. i will find solace in the fact that the current ecm direction of travel was first shown by the gem and subsequently dropped. it has a habit of doing that.

Edited February 2, 2013 by bluearmy

[Captain Shortwave]

I'm not buying an Easterly. Reasons

1) Ec op a relative outlier vs its ensembles.

2) UKMO looks ok but not on board.

3) GFS op and control not buying in one bit.

4) GEM not interested

5) ECM to bullish with easterlies all winter.

6) GFS modelled last breakdown correctly (albeit delayed) whilst ECM was showing Day after Tomorrow stuff

7) GFS got yesterday's complex system right first.

GFS has modelled Tuesday's system beautifully - super consistent for about 10 days.

corrections here

6) If GFS was right our 2 week cold spell would have lasted 2 days, it continually modelled a breakdown at 5 days. To be fair when it got to 5 days out when the cold spell actually ended, ECM and UKMO did pick up on it. Though GFS was first to spot the disruption of last fridays front which brought a snow event before milder air got in.

7) Actually UKMO had this right with the system not deepening over the south, GFS flipped flopped around and ECM backtracked 48 hours before the event after suggesting a rather nasty storm.

Edit - others got there before me and gave better responses

A little concern about GEM though, it's backed GFS throughout so far on next week. Doesn't make it right though haha. On the other hand what is going to be the big player over the next couple of weeks??? Is it the polar vortex or is it the large block which is going to form over Russia, that high doesn't exactly have anything stopping it with the polar vortex elongated through west greenland through to Alaska to the far East of Siberia. By this I mean a strong high latitude block, not specifically over the UK

Edited February 2, 2013 by Captain shortwave

[snowking]

Still awaiting the ensemble graphs, but just taken a look through the 0z ECMWF ensemble mean 2m temps across NW Europe, and the mean is around 0.5-1c warmer than the 12z suite throughout. We saw a similar change yesterday from 0z to 12z, so its worth taking a look at which is verifying best of the two at present temperature wise (the nearest we can get unfortunately is 850mb:

So, marginally the 0z ECMWF takes it in the 8-10 day range, though of course we should bear in mind this is 850mb temps, and that there is not huge statistical difference between the two - certainly not enough to say that one will definitely be more correct than the other.

As ever, it will be interesting to see if there is a swing either way in individual member post 168

SK

Edited February 2, 2013 by snowking

[Recretos]

Before you go on with the models, here are some skill scores for the near past. Basically a "percentage" like factor: 0.900 is 90% accuracy, 0.850 is like 85% accuracy, and so on. Below that, is the RMS (Root mean square) error. Basically the lower the better.

How good were the model with forecasting N. Hem. 500mb pattern, 5 days in advance.

First, this is the success of individual GFS runs. The 12E and 00E, are experimental runs. Next to it, are comparisons of specific models, GFS, ECM, UKMO, MSC (CMC-GEM) and NGP-NOGAPS. The third graph represents the comparison between GFS, ECMWF, and their control runs. In average, the ECMWF control run had better performance than the GFS operational.

Next up are ensembles. First the NCEP. Black line is the operational, red line is the control run, and the green line is the ensemble mean. This is for day 8.

And GEFS skill score "die-off" with time. It is normal for the ensembles to die-off, because as the ensemble deviation gets stronger, it is not picking up specific features that good anymore.

And ECMWF ensembles. First at day 8. Not bad I would say, and quite better than the GEFS.

And the die-off. ECMWF control, and ENS mean. At day 10, ECM ensemble mean has a skill of 0.57, while GEFS mean has a skill of 0.49.

So you will know where the models and the ensembles stand.

Best regards.

Edit: Snowking was faster with some skill scores.

Edited February 2, 2013 by Recretos

[Ian Brown]

Regarding next week , i'm not to sure GFS has it's temps rather high 6-7 deg on Wednesday . With 528 dam line covering the country , uppers between -5 and -10 .. Seems unlikely unless there expecting some strong Sunshine,

The GFS temps are gridded so the coastal influences it builds in makes it a poor guide for areas on/near the coast.

[Ian Brown]

I think with regard to Stu's points -we had a 12/13 day cold spell (in England and Wales) and the GFS had the Atlantic and Azores High coming back in at T120 on virtually every run - it had to get it right in the end ! The GEM is a poor performer IMO.

As for the northerly on Tuesday, this was always GFS speciality and is far easier to model than an easterly which will need to come within T96 before it is nailed anyway. I personally think the UKMO would go on to show what the ECM does if we could see a day 7/8 chart.

[snowking]

Well with the 0z Ensemble graphs now out, heres a comparison with the 12z from yesterday:

0z:

12z Yesterday:

Biggest differences really are less of a cold clustering on 0z vs 12z post 8-9th Feb (if you notice there was a much stronger clustering of colder runs through until around the 11th on the 12z yesterday)

However, the biggest theme still remains the huge amount of scatter, making the ensemble mean rather useless beyond 168

SK

Edited February 2, 2013 by snowking

[Tamara]

Well, I think the ECM postage stamps at t168 look pretty good to me - with a lot supporting low pressure over nearby Europe and at least something of cold ridge developing from the NE. My little eyes saw approx only about a dozen that definitively support the GFS

http://www.ecmwf.int...0200!!chart.gif

Edited February 2, 2013 by Tamara Road

[MR EXTREMES]

GFS Subjectively Observed Bias Geographical location of bias Annual/Diurnal attribute Submitted by Date Submitted Operational Implication Suspected Cause GFS too ambitious with strength and speed of systems crossing Sierras after fhr 36 SW US Cool season USGS Dec 2005 Too progressive and strong with systems crossing Sierras Model resolution of topography Convective Feedback Primarily east of front range and west of Appalachians Warm season, any time of model day NCEP HPC Spring of 1998 When specific thresholds in the mass fields are met, convective scheme is triggered and then dumps a large amount of QPF over a grid point - releasing so much latent heat over the grid point that the model is forced to adjust the mass fields by producing a local vertical motion max in the mid troposphere (~ 500mb), a corresponding upper level jet max over the vertical motion max - an intense and small scale vort max in the mid levels (MCV).

The model scales up the mesoscale circulation at mid levels and holds onto it as a real feature for as long as 3 days.

The model can produce precipitation in association with the feature as it tracks along in the flow. GFS Convective Parameterization Scheme Dry bias north of areas where over 2" of QPF has been produced in a 6hr period Primarily east of front range and west of Appalachians Warm season, any time of model day NCEP HPC Spring of 1998 QPF produced from convective feedback blocks northward advection of moisture Result of GFS Convective Parameterization Scheme QPF verification historically better than Eta CONUS Cool Season only NCEP HPC 1999 Rely more heavily on QPF from GFS - especially beyond 36 hours GDAS better than EDAS ? Aerial coverage of QPF and mass fields over done (QPF at low thresholds .01" and .10") CONUS Anytime NCEP HPC Since mid 1990's Over forecast of aerial coverage of precip can lead to high bias in PoPs Model resolution (the lower the resolution the more geography a QPF pattern can get spread over) Slightly ambitious with magnitude of high amplitude patterns North America Cool season so far NCEP HPC Since fall 2002 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 ? Ambitious to phase northern and southern stream systems in fast and spit flow patterns beyond fhr 84 North America Cool season NCEP HPC Cool season 2001 (not noticed yet in 2002) Over forecast of cyclogenesis east of Rockies Suspect related to model resolution and lack of dense obs data where associated systems originate in forecast cycle Major difference in QPF forecast than ETA CONUS Warm season NCEP HPC Since mid 1990's Lack of run to run continuity in QPF Different convective parameterizations between models result in different QPF forecasts (primarily in areas where synoptic scale forcing is weak)

http://www.hpc.ncep..../biastext.shtml

ukmo

UKMET Subjectively Observed Bias Geographical location of bias Annual/Diurnal attribute Submitted by Date Submitted Operational Implication Suspected Cause Seems better than GFS with forecast of phasing of systems in northern and southern branch of jet North American middle latitudes Anytime NCEP HPC Since fall of 2001 When GFS is showing phasing of systems beyond 84 hours, check UKMET to see if solution is consistent GDAS ?

ukmo

UKMET Subjectively Observed Bias Geographical location of bias Annual/Diurnal attribute Submitted by Date Submitted Operational Implication Suspected Cause Seems better than GFS with forecast of phasing of systems in northern and southern branch of jet North American middle latitudes Anytime NCEP HPC Since fall of 2001 When GFS is showing phasing of systems beyond 84 hours, check UKMET to see if solution is consistent GDAS ?

ecm intresting

ECMWF EPS Subjectively Observed Bias Geographical location of bias Annual/Diurnal attribute Submitted by Date Submitted Operational Implication Suspected Cause Non submitted so far . . . . . .

[Bobby]

GFS op is worse than the 00z, but looking at the ensembles the control is better and the ensembles overall look much better, here's the upper mean day 7 on the 06z vs 00z

00z:

06z:

Pattern generally shoved west which is what we want to see.

Edited February 2, 2013 by Bobby

[bluearmy]

re recretos post above

ecm control verfiying better than gfs op at day 5 !! (note there is an experimental gfs run which is outperforming the current op and matching the ecm op stat - good news. parallel soon on the way methinks)

absolutely clear that beyond approx T100, if you want to make a correct forecast of overall weather type (cold/warm/settled/changeable etc) you are better to use the ens mean, however muted the signal might become. you wont really be picking up individual features though in any meaningful way post around day 6 or 7. now you understand why forecasters use ens means in the period after day 5 as opposed to ops !

[lorenzo]

Interactive model bias page is not working just now.

Some cynical people may think that there is a reason for this...

http://www.hpc.ncep.noaa.gov/mdlbias/index.html

@Recretos do you have a link for the Korean anomalies. Not seen these online anywhere. Big Joe likes this model for sniffing out a pattern.

[Frosty.]

So we have another rubbish gfs op run (from friday 8th onwards), the 6z shows temps returning to average during friday, yes there are some rather chilly unsettled spells further into FI but also some mild days and it's generally bog standard winter fare from the gfs in FI. The Ecm continues to keep the displaced azores high out to the west and allows the cold block to strengthen and engulf the uk in very cold and snowy synoptics which look as though they would last well beyond T+240 hours.

I actually thought the gfs had woken up and smelled the coffee last night with the ecm esque 18z but it's back to it's love in with the azores high again so far today.

Edited February 2, 2013 by Frosty039

[cold snap]

ECM throwing in another epic run

Been here many times this winter , and with the gfs not on board you have to worry

Perhaps we might see the models come together in the middle

[Stu_London]

Hi Stu,

Whilst I think you are wise to keep a level head about things, I don't think many of the points you make above are really relevant here.

The first thing to bear in mind is point 6. The GFS modelled the breakdown of the last cold spell so many times that eventually it had to be correct....but its important not to ignore the fact it got it wrong so many times before! It perhaps shows the tendency we have always known that the GFS likes to revert to type. We should not assume, however, that this reverting to type will always be incorrect.

Points 5-8....you have to bear in mind that, in a strict sense, the weather from a number of weeks ago, and the associated model rights/wrongs, have very little bearing on our current potential cold spell and how long it may or may not last for. No amount of modelling will have any bearing on what the actual outcome will be, and whilst I take the point that the GFS may have modelled tuesday correctly (though I havn't verified that personally yet by having a look through the recent archives), I would argue that the UKMO was the first to be correct with yesterdays system as it never really had much interest in a particularly deep LP at any stage.

Kind Regards

SK

Hi SK

Don't get me wrong, I would welcome another cold spell as much as the next man, woman or child on the forum. However, before getting too confident, I'd like to see a few of the concerns settled.

Dont want to get into a GFS v ECM discussion, both have shown their strengths and weaknesses over the winter and everyone has a preference.

The GEM not buying in is a primary concern as this model has performed well at times this year.

And I agree the UKMO has performed very well this season however as much as it looks good and like the ECM at 144, we can't see beyond that - a few of the GFS runs have looked good at 144 - only to collapse shortly thereafter.

Slight concern that there is no word on the extended outlook from IanF - would be interesting to see if MOGREPS is anything like the ECM at days 8/9/10