Model Output Discussion - Spring Has Sprung

[Mike Poole]

ECM clusters T192-T240:

Cluster 2 (contains the op) looks interesting more as to what comes after, with low Euro heights and a ridge building west.

T264+:

Once again, promise in the extended.  Cluster 2 the puck with south Greenland high, cluster 1 with Atlantic ridge also cold.  Clusters 3 and 4 northwesterly wind for the UK, cluster 5 high retrogressing a bit.  All very jam tomorrow, but there’s wasps in the jar by now.

[stodge]

Indeed, a very poor 12Z suite for cold weather fans. After what will likely be another exceptionally mild spell this week a hint of something more unsettled later in the month and indeed perhaps a stormy end to the month with hints (no more) of negative alignment and the jet inching south.

March, we are told, goes "in like a lion and out like a lamb". Perhaps, perhaps not. The ongoing stratospheric developments do offer possibilities of a pattern change of sorts and as the PV will ultimately weaken in the month the likelihood of amplification remains.

I always thought early March had possibilities and I've not given up on that because of one poor suite three weeks or more in advance.

[RJBingham]

Nothing cold showing out to the end of the month.

 

[minus10]

Gefs 12zstill continuing with the colder conditions for last week of February..

Slightly earlier the mean going below the long term average today..towards 22nd Feb..

...its quite incredible though how different the 12z op is ..just continuing with that raging vortex to the north west..

....and now as others have stated the ecm too..

...although promise in the extended per the clusters...

...the long and winding garden path ??...

 

 

[damianslaw]

Mmm lots of change in some model output today i.e. ECM, gone are the heights lifting through the UK and instead the strength of the PV to the NW exerts control with a renewed unsettled push from the atlantic. GFS holds steadfast with the very mild south west flow becoming westerly and in time amplified.

Just musing but doesn't a SSW result in a flushing out of the westerlies so to speak before its affects imprint fully.

Many longer range modelling went for a ridge to the NW and trough to the SE, the recent suggestion of heights building through the UK didn't align with such scenario, whereas an increasingly amplified flow and eventual ridge to the west building north does.. just a few weeks later perhaps!

In the meantime the long very mild period in the reliable remains and perhaps not as settled as first thought. 

[Mike Poole]

2 minutes ago, damianslaw said:

Just musing but doesn't a SSW result in a flushing out of the westerlies so to speak before its affects imprint fully.

It does usually, when it’s on the back of a strong strat vortex, but there aren’t any significant westerlies there to flush down in this case, because of the weak strat vortex all winter, and of course the previous SSW in January.

For example, GFS 12z zonal winds:

If there were strong westerlies to flush you’d see a strong pink anomaly (right chart) in the lower strat moving down with time with the blue anomaly from the reversal above it.  It isn’t there.

[daz_4]

 Ali1977 People just don't care anymore. Chasing cold charts beyond 300 hours in middle of the February feels pointless. The time has almost run out. This is February so far + 7 days forecast.

 

[Mike Poole]

The zonal winds chart is interesting, in as much as there seems to be an increasing cluster going for a second, deeper, dip (increase in easterly winds) about a week after the initial SSW.

Will make for an interesting March, I think, and a delayed start to spring.  

[Penrith Snow]

I am here in Southern Spain sitting under Barty's big behind so to speak and the weather is glorious, 24c today and low 20's forecast all week (average 17c) locals say they have never known such a warm, dry winter so lacking in rain and we all know the reason  for that!

This month is shaping up very much like February 1998 but interestingly that ended with a notable Arctic North Westerly which brought snow well south, I remember leaving the cinema in Penrith after watching Titanic to a blizzard outside, tonight's output and the ENS hint at a cold end to the month which would likely be from the North West. I just don't see any Northern blocking any time soon despite the SSW which again does more harm than good.

I think a cold North Westerly is a credible outcome possibly leading to cold zonality in early March, winter seems a long way off but oddly so does spring.

Andy

 

[damianslaw]

 Penrith Snow Not wishing to derail the thread by comparing how synoptics have played out this winter so far against 1997-98 and there are uncanny similiarities, more so both strong El Nino years, albeit 97-98 was super el nino, a very mild December in the main which produced a chilly spell - ditto Dec 23 albeit different timescales, a rather average Jan with a mini cold spell mid month and some dry frosty weather, and now a balmy Feb as was 1998... mmm makes you wonder. 

[daz_4]

 Mike Poole Once could argue that spring has already started though.

[Mike Poole]

 daz_4 Nah!  It’s still autumn!

[ANYWEATHER]

1 hour ago, Stephen W said:

Awan I think we could well see a CET in excess of 7C in the absence of a significant cold spell (which seems likely) and continued very mild days/nights (also likely!) given the current output. We are already at 7.8C. I have to remind myself we are in February and not April !

Probably ,by April ,we will be looking at a CET closer to February's

[WYorksWeather]

I know lots of people are looking at calling things between ECM and GFS - I'd say let's hold for now.

The disagreement in the ensembles continues. ECM is much less keen on anything cold beyond 22nd, whereas GFS holds out more hope.

It will soon be obvious which is correct, so we just have to wait and not focus too much on the individual OP runs. We have very good agreement on mild up to the 20th, possibly up to the 22nd, and then beyond that we just have an immense amount of scatter. The 850hPa temperatures beyond that are ranging from 10C above average to 10C below.

[WYorksWeather]

Worth a separate post this, because I've seen a lot of people asking two things lately - how do weather models actually work, and where do errors come from?

How models work

First off, none of what I'm about to say relates to any AI-based models.

Essentially weather models should be viewed as an enormously complex physics simulation. You take in some data at a particular time, call that T+0, and then you have a series of very complex equations that relate the state of the simulation at a particular time, to the state one hour later, for example. Run this repeatedly, and you produce a forecast for the next few days up to a couple of weeks, depending on the model.

Also a good point to clear up some common misconceptions. The models don't have a built in knowledge of previous data, so for example they are able to simulate unprecedented situations that break records. They also don't generally aim to simulate certain large-scale features - they emerge from the model. For example, the MJO, weather regimes, jet stream, polar vortex are allowed to evolve freely according to the physics. The models don't have built in parameters that do things like 'promote blocking after an SSW' or 'ensure weather patterns stay close to the MJO composites'.

So, next question is where do model errors come from?

Source 1 - discretisation

The world is analogue. Ideally, with unlimited computation, you would simulate every single atom in the atmosphere and calculate its evolution. You would also like to use an extremely small time step for evolution of the model, in other words simulate from T+0 to T+0 plus one millisecond, then T+0 plus two milliseconds. Obviously that's impossible.

The way the models avoid this is to reduce the complexity by dividing Earth into a two-dimensional surface grid (typically ranging from several kilometres to several dozen kilometres on a side), dividing the atmosphere into vertical levels, and dividing time into more manageable steps (typically minutes or hours). This makes computation possible in a reasonable amount of time.

Unfortunately, it also introduces errors. Very small features, particularly those involving convection, or terrain details, are hard to simulate, because the model is too coarse to identify them. Very brief weather events may also be poorly simulated.

Source 2 - data inputs

The second type of error comes from data inputs. Infamously, this was the largest source of error in the incorrect prediction of the Great Storm in 1987. This was because data inputs were more reliant on ships, and of course with a big storm on the way, many ships were avoiding it. That limited accuracy of model projections

Of course that specific limitation is no longer present, but data inputs can still present problems. Sensors have finite resolution, can't be everywhere, and there may be delays in receiving and integrating the data into the model. This means that the model at T+0 is only an approximate view of the current state of the atmosphere, though hopefully a very accurate one. This is why ensemble forecasts are used, to make subtle alterations to the model initialisation at T+0, and see what the resulting effects are.

Source 3 - physics models

Of course, the actual physics simulations are also not perfect. Some of these are known issues. As an example from the known issues with the ECMWF forecasting model, the ECMWF does not correctly simulate the landing point of precipitation. So in cases where rain and/or snow is moving from sea to shore, the extent of inland rainfall / snow is often underestimated, as where winds are strong, the landing point of precipitation may be a significant distance from where it begins to fall at cloud level.

Another example is that the GFS has a notable upward temperature bias in certain situations. In summer, following exceptionally dry periods, the GFS can sometimes overestimate temperatures. As an example, it has at times predicted infeasibly high temperatures in the high 50s Celsius for California's Death Valley, and during the July 2022 heatwave, it predicted infeasibly high temperatures in the high 40s Celsius for parts of France.

Source 4 - chaos theory

Unfortunately, even if we achieve massive improvements in 1, 2, and 3 (and we already have, compared to decades ago), this source of error remains. Even the most trivial defects in the model tend to accumulate over time, and hence at a long lead time the atmosphere is fundamentally chaotic. Therefore, the skill of the longer range forecasts, for example the ECM extended, or the CFSv2 weeklies, is always going to be limited.

This is a fundamental limitation. It is therefore very unlikely we will ever see, for example, the ability to accurately forecast the weather in anything more than general tendencies relative to average beyond the period of 10-14 days in the UK.

Summary

Hopefully I've explained this fairly clearly. Happy to be corrected if any of the veterans here think I've got something clearly wrong or missed something out. In any case, I hope this post proves useful to someone, in which case it was worth writing.

 

 

[Rain All Night]

10 hours ago, Rain All Night said:

0z/6z deterministic runs, Mon 12th to Mon 19th (day 7)

Looks as though the GFS had it right again about those 'green' heights to our north disappearing.

0z/6z ensemble means, Mon 12th to Tue 27th (day 15)

The Scandi trough signal is still alive and kicking this morning.

12z deterministic runs, Tue 13th to Mon 19th (day 7)

Just to reiterate what I hurriedly wrote this morning, it seems that the GFS had once again been right to take a different path ahead of the other models, this time denying the possibility of significant height rises northwards of the UK.

After the passage of the next little Atlantic low through the UK, which looks like it will have its greatest impact around Friday morning, we will generally be seeing a struggle between high pressure trying to ridge up over the UK from the south, and low pressure systems from the west riding in over the top of this.

12z ensemble means, Tue 13th to Tue 27th (day 15)

From roughly 17th-21st Feb, as we're starting to see on the deterministic runs, ridging from our south may interfere with low pressure systems coming from our west. Thereafter, ridging may tend to move to our western side, while troughing may tend to occur on our eastern side - this has been the takeaway message from most ensemble suites over the past several days now.

[Severe Siberian icy blast]

 minus10 we've been saying this all winter. 

Its been an absolute disaster and I'm personally going to do alot more research into meteorology because im not convinced about some of the methods been used because in all honesty it's been a disaster from a forecasting perspective. Background signals have been misinterpreted and the significance of EAMT events working in tandem with the MJO has not been impacted like anticipated, we can point towards the strat warmings for "ruining" things but i don't buy that either. 

Its been a frustrating one to say the least but we get mugged off by the weather every single year in our country and this year has been a spectacular fail. 

But i guess im in awe of the sheer magnificence of the weather and how complicated it really is.

I want warmth now because i can't handle another false dawn! 

Edited February 12 by Severe Siberian icy blast

[Mike Poole]

Where’s the pub run going?  T276:

About time the GFS turned out a decent run…

[ICE COLD]

I know everyone has given up on this winter (including me) but the pub run offering some interest

[Met4Cast]

23 minutes ago, Severe Siberian icy blast said:

Background signals have been misinterpreted and the significance of EAMT events working in tandem with the MJO has not been impacted like anticipated, we can point towards the strat warmings for "ruining" things but i don't buy that either. 

Haven’t they? 

The recent +EAMT & more broadly strongly +ve MT event is one of the reasons we will be seeing a major SSW next week. 

Rossby wave packets generated in part due to momentum being deposited directly into the Pacific jet & causing the jet to extend eastwards have allowed for significant heat flux into the stratosphere to finally deliver the knock out punch. Indeed, Tamara mentioned this likelihood a couple of weeks ago and now we’re seeing it play out. 

The strongly +MT was signalled quite a while ago as the MJO progressed through the Maritimes & into the Pacific generating strong +FT, MT always follows FT. 

In terms of the MJO not impacting as expected, the above is an example of an anticipated consequence of the MJO, the reason we’re not currently seeing blocking despite favourable MJO is because the MJO cannot & should not be used in isolation. Here’s something I posted back on January 22nd: 

Quote

What did previous look like a good chance of sustained high latitude blocking during February is rapidly fading. 

Despite an increasingly favourable MJO signal into the western Pacific, we’re missing a vital ingredient & that’ poleward propagating +AAM (westerly momentum). 

Unfortunately, this time +AAM is increasingly glued to the equator & we’re not seeing this fluxing towards the pole, this helps (despite the MJO) to inflate the European ridge & indeed, a very strong signal for this within NWP through the remainder of January. A flat, mostly mild end to the month with low pressure systems deflected northwards, potentially stormy at times for Scotland. 

So actually, this has been rather well sign posted for a couple of weeks now & those of us that subscribe & try to apply the GSDM framework have spoken about this likelihood a few times. 

Edited February 12 by Met4Cast

[Uncertainty]

The night is always darkest

Just before the dawn!

It took a pretty crazy storm to get there @Mike Poole - as you say s’about time we had some excitement. It’s been dismal in here recently  - recriminations galore (the vast majority of it completely understandable!) after our 351st fail of this utter let down of a winter.

The pub run shows that a) another potentially hyper stormy period is plauisble and b) Atlantic blocking is not off the table but is unlikely until over 12 days away.

 It does turn into a very cold and snowy run

[Ice Day]

Well it's been a while since we had a pun run special, but it's churning out a corker at day 13!

 

Very cold and an unstable airflow, very nice.  However, this has about as much chance as any other day 13 pub run chart 

[Allseasons-Si]

Evening,...just got in and i am liking the gfs 18z,...BOSH

and cold enough for snow to all levels.

it's the crazy pub run antics but i will take it.

[Tim Bland]

Let’s hope we can at least go out with a bang after this terrible winter ! 

[Jordan S]

Purely from an observation point of view than a forecast in regards to this low below but a rather St Jude (2013) looking storm system showing this evening on Gfs, not on par with intensity as the low exits the UK though and it’s not at time scale to take too literally but shows the lows track and timeframe itself ties in with recent updates on potential snow on northern flanks of systems moving through within a northwest airstream as we head towards months end, though the signal for heights over Greenland and Iceland have weakened so that’s the only real difference.

 

The Synoptics Gfs showing this evening towards end of February is not unexpected, rather was expected even if conditions will probably be slightly less amplified come the week itself, the low itself above unlikely to be there on the day shown ofcourse, anyway from a lover of most weather types, plenty to be intrigued on for the second half of February with slight convective weather and snow still possibly being involved to some degree for the usual spots anyway later next week but that’s without significant frontal rainfall in the flow which would ofcourse be likely to bring snow levels down to where more are in a place to witness the wintry stuff, though probably not the south. All fast paced stuff like a skipping rope has been waved across the North Atlantic shaking the stagnant synoptics pattern up a bit so to speak.

 

Will give a proper update by the weekend with a couple of different charts than the usual