Model Output Discussion - cold spell to end Autumn

[Frosty.]

7 minutes ago, Ed Stone said:

As lovely as those fax charts are, Karl, they beg an awkward question: why are eastern parts not included in the MetO's warnings?

Good question Ed, i think its because the met office expect the flow is initially more west of north so exposed western and northern areas are in the sweet spot but the fax charts indicate the flow veering northerly so eastern counties bordering the north sea, including the southeast would then see a big increase in snow showers.

Edited December 5, 2017 by Frosty.

[karlos1983]

3 minutes ago, Paul_1978 said:

People say this is no 2010, which is quite correct, but surely we are in a prime position for cold and snow over the coming weeks, if perhaps not as severe as that famous winter.

So don't be disheartened. To be honest, if we can get a repeat of the events Dec 1990 and Feb 1991 alone then that would be a great winter!

So whilst I won't be looking for a 2010, I'll certainly be looking for a one off "big" snow event. Surely we've got a higher chance than normal this winter. 

Agree with that Paul. I prefer to say this is no 2014/15/16, 2010 was just ridiculous and if one of them pops up again then hooray!!! but it's probably quite unlikely.  

[Duane S.]

10 minutes ago, Craig84 said:

Hi everyone.

I've been reading this forum a good few years now and I'd like to say thanks to all the regular posters who contribute to this great place for weather geeks!

Thought I was about time I contributed!

Im terms of potential for later in the month, I'm really encouraged by the 06z GEFS. 80% are attempting to build heights NE by the 20th. Shown by the mean charts below

 

 

Way into FI but it seems a strong signal and one to watch.

 

Welcome to the madhouse Craig.

Definitely one to watch, the signal does appear to be gradually growing .

It does appear at the moment, and through most of December, we will have a continuation of the current status quo. A Mid Atlantic ridge waxing and waning (the degrees to which will likely determine whether cool / cold / very cold, dry / rain / snow here) which would likely negate any major influence from the east. Until perhaps right towards month end and into January when Pacific forcing may be more favourable for heights to the NE to start becoming a major player in events.

Not too bad shabby in the meantime mind.

[Methuselah]

7 minutes ago, Frosty. said:

Good question Ed, i think its because the met office expect the flow is initially more west of north so exposed western and northern areas are in the sweet spot but the fax charts indicate the flow veering northerly so eastern counties bordering the north sea, including the southeast would then see a big increase in snow showers.

Here's hoping!

[Devonshire]

6 minutes ago, Barking_Mad said:

Thanks for this. I struggle to get my head round this in a flat top down view as it's just a 1D image. How does all this look if you were to draw it from a side on 2D view? 

You appear to be missing a dimension BM! I also sometimes wonder whether a 3-d representation (in 2-d of course - not talking holograms) of model output is available (to the public) - it would make things a lot clearer to me (possibly). A 3-d representation of a slider would make an interesting video - does anyone know of one?

[Southender]

5 minutes ago, Devonshire said:

You appear to be missing a dimension BM! I also sometimes wonder whether a 3-d representation (in 2-d of course - not talking holograms) of model output is available (to the public) - it would make things a lot clearer to me (possibly). A 3-d representation of a slider would make an interesting video - does anyone know of one?

Something like this?

http://www.meteociel.fr/modeles/gfse_3d.php

[ArHu3]

6 minutes ago, Devonshire said:

You appear to be missing a dimension BM! I also sometimes wonder whether a 3-d representation (in 2-d of course - not talking holograms) of model output is available (to the public) - it would make things a lot clearer to me (possibly). A 3-d representation of a slider would make an interesting video - does anyone know of one?

 

Something like this? 

http://www.meteociel.fr/modeles/gfse_3d.php

 

There was another where you can look at the different layers but I don't know the link

[MattStoke]

People commentjng as if the ECM is spot on with the slider low. You really think a model will be spot on with such a complicated setup this far out? Ridiculous! It will be days before that one is resolved. The huge difference in the handling of it by the various models over the past few days should tell you that. Come on, people!

[Methuselah]

1 minute ago, MattStoke said:

People commentjng as if the ECM is spot on with the slider low. You really think a model will be spot on with such a complicated setup this far out? Ridiculous! It will be days before that one is resolved. The huge difference in the handling of it by the various models over the past few days should tell you that. Come on, people!

Not to mention that (and the Mystic Megs ought to know this.) it's still 5 days' before it even happens!

[MattStoke]

Mind you. The ECM has over half a metre of snow here by next Thursday. So I should really be singing its praises 

[Captain Shortwave]

One can hope that the NAVGEM has nailed this one

If low pressure can get steered somewhere towards the English channel then near enough everyone under the frontal system should get snow. We just need the forcing from that weak area of higher 500mb heights to steer that system south east and not through the UK.

This is what we are looking for from the 12z suite later. 

Edited December 5, 2017 by Captain Shortwave
Fixed lassie23 :)

[Frosty.]

A lot has been said about the progressive Ecm 12z operational last night and again this morning but last nights ensemble mean was a peach for wintry conditions going forward and looking at this mornings 00z..I really can't see a downgrade..we have an arctic blast looming and beyond that this shows generally cold unsettled weather with a risk of snow, especially in hills but with wintry ppn in many areas at times and continued risk of overnight frosts and ice...I'm not seeing a change back to anything mild / zonal going by this! 

Edited December 5, 2017 by Frosty.

[lassie23]

Just now, Captain Shortwave said:

One can hope that the NAVGEM has nailed this one

<img src='http://images.meteociel.fr/im/9901/tempresult_skd2.gif' border=0>

can't open the link

[bluearmy]

9 minutes ago, MattStoke said:

People commentjng as if the ECM is spot on with the slider low. You really think a model will be spot on with such a complicated setup this far out? Ridiculous! It will be days before that one is resolved. The huge difference in the handling of it by the various models over the past few days should tell you that. Come on, people!

It won't be days though. The envelope will get smaller from now on. By tomorrow's 12z runs we will know the area at risk of a slider or if we will see a front just move through with snow on its front edge and potentially all snow in the north. Clearly the slider scenario is more tricky to pin down the band in the firing line but because we don't have a block in place to our east, it should be a little more predictable re easterly progress. 

[bluearmy]

1 hour ago, Man With Beard said:

 

... and back to the one cluster FI, which persists with a shallow area of heights to the north:

 

Looking at the eps probability maps, it's surely not conceivable that there is only one cluster. 

[Sky Full]

2 hours ago, Daniel* said:

One thing for certain somewhere in the country, is going to get a pasting. Place your bets..

Pembrokeshire.......

Its NOT you!  

 

[johnholmes]

re this comment

Maybe the experts could refresh us on how sliders are modelled in regards to their biases? This might give us a hint of what to expect in a few days? 

No model has any bias, all are run with the mathematics to deal with the laws of physics as applied in meteorology. Lows and highs appear/disappear on this basis.

At the risk of causing some dissent another point worth trying to remember. That is compare 00z to 00z and 12z to 12z, not the most recent, that is until the time period gets below, usually, about T=96. Around this time then the more recent data increasingly tends to outpoint this method. 

Edited December 5, 2017 by johnholmes

[Frosty.]

8 minutes ago, johnholmes said:

Lows and highs appear/disappear on this basis.

There's plenty of lows and highs in here too during a Gfs / Ecm run commentary..all good fun

Anyway, the models show it turning very cold from the north with blustery snow showers to the west, north and east as winter sinks its icy teeth in during thurs, fri and the weekend.

[Hocus Pocus]

7 minutes ago, johnholmes said:

re this comment

Maybe the experts could refresh us on how sliders are modelled in regards to their biases? This might give us a hint of what to expect in a few days? 

No model has any bias, all are run with the mathematics to deal with the laws of physics as applied in meteorology. Lows and highs appear/disappear on this basis.

Indeed John. Though for some the laws of physics need not apply when viewing any model output when it shows something they’re not too keen on.

From years off model watching purely from a rank amateurs perspective these sort of set ups are never resolved until +24-48hrs, the genral theme of a low dropping on a SE axis is relatively straight forward once within these timescales but the boundary between rain and snow less so with so many variables that come into play. A great learning curve watching this unfold though.

[SnowBallz]

3 minutes ago, johnholmes said:

re this comment

Maybe the experts could refresh us on how sliders are modelled in regards to their biases? This might give us a hint of what to expect in a few days? 

No model has any bias, all are run with the mathematics to deal with the laws of physics as applied in meteorology. Lows and highs appear/disappear on this basis.

That isn't strictly true, John.  Whilst you are correct in respect to how thermodynamics are considered primary drivers in NWP, various other factors and variables need to be introduced to account for regional variation and idiosyncrasies, in order to improve accuracy.  The difficulty, is that the layering of variables increases error rate by an order of magnitude, and therefore models require tuning.  It has been observed many times over that some NWP models are predisposed to resolving certain patterns or outputs more favorably than others, and there is no real explanation other than how the underling mathematical weighting has been written.  Does that constitute a bias?  In terms of its output, possibly - but it is entirely intentional, and in any event anomalously recurring patterns are generally synthesized out (ensembles are particularly sensitive to this)

SB

[johnholmes]

3 minutes ago, SnowBallz said:

That isn't strictly true, John.  Whilst you are correct in respect to how thermodynamics are considered primary drivers in NWP, various other factors and variables need to be introduced to account for regional variation and idiosyncrasies, in order to improve accuracy.  The difficulty, is that the layering of variables increases error rate by an order of magnitude, and therefore models require tuning.  It has been observed many times over that some NWP models are predisposed to resolving certain patterns or outputs more favorably than others, and there is no real explanation other than how the underling mathematical weighting has been written.  Does that constitute a bias?  In terms of its output, possibly - but it is entirely intentional, and in any event anomalously recurring patterns are generally synthesized out (ensembles are particularly sensitive to this)

SB

We could discuss the pros and cons of how we each see the laws of thermodynamics perhaps until the cows come home. Essentially I suspect we are both saying the same thing but in a slightly different way.

[bluearmy]

We know for certain that models have bias’. NOAA forecasters often reference the gfs/gefs ones when making their text assessments. They surely not intentional but they exist within the model.

as snowballz comments above, trying to rewrite the algorithms to avoid them would be a bigger job than just taking them into account when assessing the output. 

Joe B on weatherbell often shows the bias of the ec46 to bring the trough into the western coast of the USA. He says it’s to do with feedback loops because of the rockies but whatever it is, it happens quite a lot. 

[Methuselah]

TBH though, Joe B could arguably do with having a few of his own algorithms rewritten?

[Guest]

MORE ON SLIDERS AND WHAT TO LOOK OUT FOR

Further to my post late last night describing a slider event which I experienced back on January 9^th 1968 (now buried on page 151) I see that there continues to be considerable confusion as to how a slider behaves. Now, whilst I am certainly not a technical expert, I do have considerable experience of slider events and I must have seen well over 100 of them in the last 55 years since I started studying weather maps every day. This included over 10 of them in the epic 1962-63 winter alone (which I described in my post on page 134).  I shall endeavour to demonstrate what to look for once any model run has shown one developing on their charts. I shall use the most current GFS 6z run as an example which is still churning out whilst I am writing this post.

Now I fully appreciate that next week’s sliders projected on most of the recent model runs may not materialise at all. It is possible that the LPs around the south of Greenland or in the mid-Atlantic may not break away and slide south-east and they can easily deepen and take a much more northerly route from west to east or even south-west to north-east – ie: not a slider event at all, with much milder conditions spilling across most or all of the UK. At this stage, this appears to be the least likely scenario, fortunately for the coldies on here. Even those models which show this flatter and milder pattern for next week generally return to renewed mid-Atlantic amplification and colder conditions later on in the run.

Some runs have shown successive slider events and others just a single event. Almost every run is at least slightly different with varying routes for the slider and these changes should be expected right up to T+24. This is not really due to major changes, although the models are generally struggling to nail down some of the broader pattern changes, let alone any detail. It’s because just a tiny change in exactly where the LP breaks away can make a big difference on the path that it takes. With the jet stream also changing direction slightly this can influence the path. Just minor changes in the position and strength of HPs and ridges adjacent to the LP can also impact on any changes down the line.

So, to keep things simple (for me as well as many of the less experienced readers), I shall pick things up from where one of these sliders has been predicted on the current GFS 6z model run (I started writing this 1130). First of all, let’s focus on the run from T+108 through to T+144 starting off with the surface pressure charts:

GFS 6z, December 5^th -  Surface Pressure Charts:

                            T+108                                                 T+114                                                 T+120        

                               

                                      T+126                                                T+132                                                 T+144

                               

Going through this run you can see the slider breaking away from the LP south of Greenland and taking an east-south-easterly path, reaching Northern Ireland by T+120 and deepening only very slightly. It continues on this route and crosses northern England and has moved out into the North Sea by T+132 having continued to deepen very slightly. At this stage, it looks like much of central and southern England will be under the milder (or less cold) air to the south while Scotland should get some heavy snowfall (see later).  Now if we jump to T+144 the slider phases with the larger LP over Scandinavia and it deepens a little more and engages with some much colder air there and sweeps this across the UK in its rear. Having done its job as a slider, it then changes its behaviour and moves south into the southern North Sea maintaining central pressure.

Now let’s look at what is at least partially responsible for these changes, the meandering and buckling jet stream.

GFS 6z, December 5^th - Jet Stream Charts:

                           T+108                                                   T+114                                                  T+120    

                                  

 

                            T+126                                                  T+132                                                  T+144

                                       

On this run at T+108, the GFS shows a branch of the jet stream running south-east from Greenland, across Iceland and through the UK. This is already being pushed away eastwards by a more powerful and more direct west to east flow seemingly set to blast right across the UK. By T+120 the jet is already showing signs of buckling, weakening slightly and moving further south. Remember, all the time, the slider LP is on the northern and cold side of the jet stream. The remnants of the jet that ran south-east from Greenland despite fragmenting and weakening are still having an impact in helping to push the main stream further south (it’s much more complicated than this not least because we are dealing with a 3 dimensional jet profile but I’m keeping it simple so that we can see what impacts the small changes in strength and direction of the jet have on the pressure patterns). From T+120 through to T+144 the jet has buckled further and, although broader and stronger, it’s meandering favourably for the UK now pushing well to the south of us. By the time the slider low has phased with the LP over Scandinavia, this allows the whole system to move southwards.

Next I move on to the 850 temperatures and I feel that these can give some of biggest clues to a slider's behaviour - this seems to be quite widely misinterpreted.

GFS 6z, December 5^th- 850s Temperature Charts:

                             T+108                                              T+114                                                      T+120

                                 

   

                             T+126                                              T+132                                                      T+144

                                

We see that at T+108 the UK is widely under -8s with -4s to -6s further west. It looks like the area of 0s to +4s or even higher is about to push right across the UK but this is misleading. Focus on the narrow strand of +4s looping to the north-west. This is in the area where the slider LP is just beginning to breakaway east-south-eastwards. Look at the shape of these positive 850s, the northern arm is actually being lifted or squeezed out. By T+114 that northern strand is fragmenting and by T+120 you can see it being pushed away south and south-eastwards to the extent that in only 12 hours just southern England is likely to see positive 850s rather than the whole country. The slider is doing its job. It “does not engage with the mild air to the south” whilst it remains on this east-south-easterly track. It is actually starting to engage with the colder air ahead of it to its north and east. The only way this could change would be if the jet stream takes a more northerly route pushing the slider LP on to a more north-easterly path, in which case it would no longer be a slider and would take on the properties of a more typical Atlantic LP sucking increasingly milder air into its circulation.

As we have seen above, GFS project the jet stream to take a more southerly route. So what looks like a bad evolution for coldies is actually a very good one – perhaps a few of those posters who view each chart as they are churned out and start to moan about the apparent warmth about to swamp the UK can take note of this. Waiting for at least the next few charts and referring to the changes in the jet stream might lead some to change their view very quickly!  I say this in relation to this type of set up but it equally applies to any situation where a broader pattern change is indicated and seems likely.

Moving on to T+126 we see how far south of the UK the above 0c 850s have already been pushed. This shows that the segment of mild air will not get caught up in the slider system. Yes, it will drag the less cold air across the south but unlike a more typical Atlantic LP there will be very little mixing in the centre of the slider. It’s not all good news “at this stage” as we are left with mostly -2s to -4s over much of the UK with only the far north of Scotland under sub -8s. What we can see is that there is now a much wider area of sub -2c 850s not only ahead of the slider but also coming in from the west behind it. The reason why we only have extremely marginal 850s at this stage is due to the slider taking an “east-south-easterly” path – just a very small change to a more direct south-easterly path would maintain as well as suck in rather lower 850s (sub -4s or lower) from the north and east. Alternatively if the whole system was just a little further south (eg: taking a track say through the Midlands) a lot more cold air would be retained throughout with sub -6s and lower more generally (except in the far south). Please note that I am going by the 850s that GFS have been showing on this run which may actually be slightly higher or lower than those predicted even if this precise pattern verifies (pretty unlikely).

Going through T+144, we see that with the LP now phased with the LP over Scandinavia, that it’s starting to drag in rather colder air behind it – nothing remarkable at this stage but going in the right direction. A larger part of the UK is once again under -4s and there is a pool of sub -6s developing over north-west England. Remember some of these uppers are now being sourced from Scandinavia and north-west Europe with a much more continental influence. This will likely engage drier air with much lower dew points as well as far less modification from warmer seas. This should gradually result in favouring most marginal precipitation events to fall as snow with only around -4c 850s (or even slightly higher). Looking forward to T+192 (not shown) it looks like GFS wants to rinse and repeat the pattern.

Now a quick look at the surface temperatures for the same period.

GFS 6z, December 5^th - 2m Surface Temperature Charts:

                            T+108                                                   T+114                                                  T+120        

                                            

 

                             T+126                                                  T+132                                                 T+144

                                   

We start off with temperatures over most of the mainland UK between 0c and -4c. The milder air only gets into the far south and far west for just a few hours with up to 8c there. Most of England is closer to 4c and the far north and much of Scotland remain closer to 0c throughout. Towards the end of this period most parts are closer to 0c with night frosts returning. These values (+ the 850s) will be reflected in which areas get snow rather rain or sleet.

GFS 6z, December 5^th - Precipitation Charts:

                                        T+108                                                  T+114                                                 T+120   

                                     

 

                              T+126                                                  T+132                                                 T+144

                                     

Now I selected this particular slider scenario as it doesn't really produce much in the way of snowfall at this stage of the run. I could have chosen one from another run which takes a path further south and leads to much greater snowfall but I did not wish to mislead anyone. The purpose of this exercise has been to point out the basics of what to look for and what to follow with any of these slider scenarios. The charts above start off at T+114 with the legacy of the earlier northerly and some snow showers in parts of the country. There is some snow in the east and north ahead of the slider with rain further south and west. This does give an idea of how much more wintry the precipitation might be with a very minor shift southwards in the pattern. By T+126 most of any snow in the east has turned to rain but there are significant snowfalls indicated for the far north of England and Scotland. Then look at what is going on further north. The slider is engaging with the colder air on its northern flank and dragging it southwards again in its rear. There is some widespread and quite heavy snowfall to the north of Scotland and the northern isles would be prone to large accumulations in this particular scenario. By T+132 the area of snow over Scotland is pushing southwards. The rain further south is clearing away too. By T+144 snow showers are shown down exposed west, north and east coasts.

GFS 6z, December 5^th - Snowfall Accumulation Charts:

                              T+108                                                  T+114                                                T+120    

                                    

 

                              T+126                                                   T+132                                               T+144

                                     

Snow accumulation charts are notoriously unreliable and this exercise actually reveals some of this. We start off with the residual accumulations shown from any snow from the earlier northerly. Given the showery nature of that air stream with possible disturbances and troughs probably embedded in it, possible larger accumulations might have been seen in some inland areas away from the more favoured exposed coasts. So, the starting position is almost certainly pretty inaccurate. Let's focus on the changes shown as the slider passes through. At T+120  around 2 cm + is shown in England and Wales across all but the far south with greater accumulations further north. Based on the path of the slider, I feel that this area of snow ahead of it, is shown rather too far south, although it is possible that there is a short period of snow over a wider area prior to it turning to rain. Much of any lying snow is shown to melt later in the period but only in the Midland southwards. Meanwhile the snow cover further north is shown to deepen. There is no point in my analysing the snowfall any further on one particular run around 4 to 6 days out. Even if this model run comes anywhere close to verification the marginal temperatures and the intensity of the precipitation will probably not even be certain in the nowcasts! 

Now remember that this is just one example of a great many scenarios. If we do see a slider it could take any path across the UK, although given the ”usual” south-east trajectories, I would favour somewhere between the Scottish borders and the English Channel. Some sliders move too far south and miss the UK altogether and move into France.

For those of you who want to view these changes in more detail and look for other potential sliders and haven’t done this before, here is the link to the Meteoceil charts:

http://www.meteociel.fr/modeles/gfse_cartes.php

I hope that some of you have found this very simple analysis useful and will have a slightly better handle on what to look out for.

I think that I have said enough about this subject for now. I am in real danger of being dubbed “Mr Slider” with my obsession in the same way that @nick sussex is dubbed “Mr Shortwave”!

 

[radiohead]

1 minute ago, Ed Stone said:

TBH though, Joe B could arguably do with having a few of his own algorithms rewritten?

Whatever about model bias, humans certainly have their own bias!