Model output discussion - January cold spell ending, what next?

[Penrith Snow]

57 minutes ago, ThamesStreamer said:

Christ, look at the switch in the GEFS from the 00z (left) to the 06z (right). Must be something afoot to see such a pronounced movement in the ensembles from a crazy spread to a more confident outlook?

Christ indeed, even the good Lord must think the long suffering members on Netweather deserve a break. 

Its best to ignore single Stella runs like the 06z GFS para but you can't ignore such a clear switch in the ensembles which to be honest is remarkable.

Reminds me of mid February 2009 when the ensembles suddenly switched from very cold although with some scatter to a clear signal for mild. Sure enough mild won.

Andy

[IDO]

1 minute ago, Cold Winter Night said:

The verification process of these models is a little bit more sophisticated than simply eyeballing a one month line graph of just one variable (500hPa heights). Tomorrow things might look different again.
Based on these data alone, the only thing you could say is that they are all extremely close.

Add to that the fact that we are currently in a situation with rare events and combinations of background signals and this might become a good test for the verification of all models in a complicated setup that includes both Troposhere and Stratosphere.

These have a 30-day sampling so have few issues in the context of a fair representation. 

Heights are a good choice as they pertain to amplification deviance and I have looked into when the ecm performs badly in the individual daily drops and they mostly relate to overdoing heights.

It is not the average over the 30-days per se, as you say relatively small differences, it is in the dips in verification which drags the average down from gfsP>gfs at d5, to gfs>gfsP at d10 (also ecm). Look for the dips and extrapolate that to the actual charts and generally the causal link is excessive height estimation by ecm (and I suspect gfsP).

eg 15 Dec d10 chart verified at 0.26 compared to the 30-day mean of 0.561 (ecm):

The d10 forecast and the actual day:

The forecast was for two plumes of heights, the reality was more trough dominated. Going from a UK HP to UK LP! Those two ridges much less amplified than the 10-day charts suggested.

[Vikos]

Again, those EC Scenarios are

[sheikhy]

2 hours ago, bluearmy said:

I currently have medium term starting at day 4 !  I do take your point ..... tbh, the output day 3/4 to day 6/7 is just noise at the moment.  Unreliable and nothing nice to look at in any case! 

Ecmwf 06z ever so slightly further west than the gfs 06z at 72 hours with the colder uppers!!!!assume the front is further west as well granted only a few miles maybe?lol

[Frosty.]

Oh man, this place is an insane asylum.. but here is the GEFS 6z mean..bank, just bank already!!!!  

 

Edited January 10, 2021 by Jon Snow

[Timmytour]

3 hours ago, Nick F said:

Yes, an Atlantic frontal system moving in from the west, bumping into a retreating cold high ahead of it over the near continent. Snow to rain event. Actually GFS similarly has the frontal system and developing low along it dropping SE into France, but without the preceding snow. May not be replicated on following runs though.

A la a 1979 forecast?

[Frosty.]

5 minutes ago, Timmytour said:

A la a 1979 forecast?

1979 will mean diddly squat to most new members on here..what happened in 1979??.hmmm..on second thoughts, don’t bother..

Edited January 10, 2021 by Jon Snow

[bluearmy]

10 minutes ago, sheikhy said:

Ecmwf 06z ever so slightly further west than the gfs 06z at 72 hours with the colder uppers!!!!assume the front is further west as well granted only a few miles maybe?lol

It makes a diff to far ne eng and e Scot !  Certainly seeing a backing west of the front though exactly how big the swing can still be is open to question ....

[Man With Beard]

1 hour ago, Battleground Snow said:

I'm going to put a positive slant on its and say because it verifying better than the OP at 120, it has the cut off Low nailed 

Has anyone got the stats that include the reliability of the ensemble means? I recall seeing them a couple of years ago and the ECM ensemble mean beat all other runs from D8 onwards?

[Trom]

1 hour ago, Met4Cast said:

As some have already pointed out, the GEFS 06z show a huge shift towards a colder outlook compared with previous runs. Not sure when the last time such a dramatic shift occurred, but 2010 stands out in my mind.

GFS Op is a clear outlier now. I would also put more weight in the GFS(P) simply because it has more vertical layers so should resolve any SSW downwelling more accurately. 

After a tricky few days, things are steadily improving towards a colder outlook for the UK. EPS means/clusters are strengthening the signal with each model run. 

We're reaching the "tipping point" where we start to see increasingly cold and blocked runs appearing in the mid-extended. Buckle up!

I can see the argument that the GFS(P) has more vertical levels and theoretically should be able to model the events in the strat more accurately than the existing GFS.  However, it's worth remembering that even the parallel has marginally fewer levels than the ECM.  So for those hanging their coats on the parallel, I'd be looking for some agreement from the ECM before climbing aboard. 

[BalderstoneNick]

Sorry if this is common knowledge, but just for anyone who wished there was more ECM data to view, you can by simply going on the ECMWF website, they decided last year that more info would be freely available to the public:

ECMWF | Charts

APPS.ECMWF.INT

 

[nick sussex]

The JMA 06 hrs run at T84hrs has also shifted west and has a shortwave waiting to head se like the GFS .

Given the timeframes involved we should know by this evening whether the new trend is correct .

[Yorlum11]

13 minutes ago, Jon Snow said:

1979 will mean diddly squat to most new members on here..what happened in 1979??.hmmm..on second thoughts, don’t bother..

I think its similar to the 1837 forecast. Just clearing that up for you 

[Methuselah]

A fair amount of scatter in today's 06Z temperature ens... Though the operational is not an outlier, it is substantially warmer then most other members:

    

And the NH profile is still headed in the desired direction:

    

    

As an aside, does anyone know of any verification stats, that pertain to SSWs?

 

Edited January 10, 2021 by General Cluster

[Timmytour]

2 hours ago, Tim Bland said:

X rated

Nice scan...hope the baby looks that good after the birth!

[IDO]

18 minutes ago, Man Without Beard said:

Has anyone got the stats that include the reliability of the ensemble means? I recall seeing them a couple of years ago and the ECM ensemble mean beat all other runs from D8 onwards?

They stopped the reanalysis of the mean in 2019 sadly, but yes the mean was far more reliable post-d8.

[wellington boot]

1 hour ago, Waterspout said:

The v16 code was actually signed off and approved back in May 2020 by the Ops Director. This Parallel running October to the end of this month is purely “IT” related. The actual verification improvements were assessed by EMC based on the previous year’s initialisation data to evaluate performance compared to v15.

They won’t pull the code on a project that started back in 2019, on those Stats you published for the 3 week period you illustrated. This is the new and improved code, even if it’s an overall 1% increase, it’s still improved after extensive testing on long term previous starting data spanning a longer period than those weeks you highlighted.

I didn't actually offer any statistics.

But you are right. I meant to add one other comment, which was that verification over the past month might not match previous stats and they must have been monitoring for much much longer than just the past month...

[East_England_Stormchaser91]

Morning all again. You’d think the runs would be a bit more sobering, but they really aren’t! 

I do think we are overdue a Feb 1991 or January 1979 style cold spell, and have not seen an opportunity like this so early on in the year for some time. Maybe the first time ever since setting eyes on these models. The BFTE started this way but of course it was mid February by the time the fun charts were coming out. 

[BARRY]

32 minutes ago, Timmytour said:

A la a 1979 forecast?

twas a bloody cold winter i know was working on a 21ft open all yesr round ferry

[Timmytour]

24 minutes ago, Jon Snow said:

1979 will mean diddly squat to most new members on here..what happened in 1979??.hmmm..on second thoughts, don’t bother..

LOL.... possibly my favourite experienced weather event! TV weather showed  westerly winds spreading east melting the snow that had fallen. I went to bed with a heavy heart. listening to the dripping of melting snow.  When i woke up there was six inches of snow...overnight the progression had been halted in its tracks and pushed back.  I'm hoping what we are seeing  in the models will provide that same impetus to push back anything approaching from the west.   

[phil nw.]

Just flicking through the 06z runs and ens. set and there does seem to be a move in the medium term towards the coming cold spell.

First a and  look at day 8 charts from the 06z GFS runs incl GFS/GFSp/GFS mean.GFSp nearer to the mean chart i would suggest with the evolution of better ridging in the Atlantic and the angle of approach of the Scandi.trough from the ne.

Comparing the 06z gefs diagram from yesterday (first image) and we can see more of a colder dip by day 5 which would tie in with the cold air perhaps not clearing the UK easily as the Atlantic hits the buffers somewhat.

As shown in the 500 images the 06z op shows as a warm outlier from the key mid-term period.

Whilst i am scanning the outputs i will just add the 00z ECM mean 192 chart and it's ens diagram too

Short term uncertainties aside it seems the ongoing 500 hPa  pattern of Atlantic ridging and Scandi.troughing developing around day 8 is still very much the favoured trend.

Taken all in all then from all these charts and graphs we can see a steady build up of the colder evidence and also with the timing around 18-20th.

 

[goosey007]

Hi all,

great analysis in here over the past few weeks! Not often I reach out but as a novice in the interpretation of the models can someone sum up in leymans terms and explain to me what is going to happen around the 20th Jan ? I am just confused by the models ....and the GFSP is adding to it!

much appreciated!

D

[phil nw.]

Quite a few posts have been moved here guys

Just a reminder just keep to current outputs and discussion in here please,Thanks.

[MATTWOLVES 3]

I will just put this out there regarding ensemble forecasting, and the pros and cons associated with it.. It may be of benefit to those who are new or uncertain... And to some who hang onto each and every run for dear life. 

An ensemble forecast is a set of forecasts that present the range of future weather scenarios. Multiple simulations are run, each with a slight variation of its initial conditions and with slightly perturbed weather models. These variations represent the inevitable uncertainty in the initial conditions and approximations in the models. 

The uncertainty associated with every forecast means that different scenarios are possible, and the forecast should reflect that. Single deterministic forecasts can be misleading as they fail to provide this information. 

By generating a range of possible outcomes, the method can show how likely different scenarios are in the days ahead,and how long into the future the forecasts are useful. The smaller the range of predicted outcomes, the sharper the forecast is said to be. Good ensemble forecasts are not just as sharp as possible but also reliable. If a reliable forecast says that there is a 70% chance of top temps rising above a certain threshold, then in 70% of cases when such a forecast is made temperatures will indeed rise above that threshold. 

Also to take into consideration is our lack of knowledge does significantly increase uncertainty in the forecast. This is why there is much work going into improving our knowledge of initial conditions and of atmospheric processes that computer models need to mirror. In addition, the atmosphere is a chaotic system. This means that it is sensitively dependant on initial conditions. In a chaotic system, a slight change in the input conditions can lead to a significant change in the output forecast. In a non non chaotic system, small differences in initial conditions only give small differences in output. Hence, it is important in forecasting to investigate how sensetive the atmosphere is at any stage to initial conditions. Ensemble forecasting does this by looking at a spread of outcomes. 

An ECMWF produces 51 forecasts. On of these is known as the control forecast and is produced with the best available data and unperturbed models. The other 50 are produced with slight perturbations of initial conditions and of the models. Taken together the forecasts become the ensemble, which shows the range of outcomes and the likelihood of their occurrence. 

So in a nutshell Human input and its partial lack of understanding, can have an affect on some of these ensemble forecasts, and it certainly doesn't pay to get to down hearted at the sight of 1 or 2 poor operational runs... And with what is going on with the strat right now, only serves to make it even more of a dilemma for such models. 

Information was sourced from the main ECMWF data base. 

Edited January 10, 2021 by MATTWOLVES

[Daniel*]

The Siberian express is returning the question is will it reach its final destination? (UK).

 

GEFS are now also aboard something is afoot that is clear. For coldies we have had to wait for 3 years for any shot of proper cold spell, but it seems our patience could be rewarded.