Model output discussion - 5th December 2016 - Into Winter we go

[Gael_Force]

10 minutes ago, bluearmy said:

Can we just agree that a mean temp anomoly that is for more than a month is going to be pretty worthless anyway ??

if you can see the individual monthlies within the period then you can make an analysis of sorts but even that is fraught with likely issues. 

 

The only LRF output, of those we have access to, I have seen get the OND quarter nearly right is the Huug Van den Dool SSTA analogue forecast. The prediction back in August.

[Seasonality]

50 minutes ago, Ravelin said:

Since it's 'quiet time' between models runs I have a question. On the GFS ensembles graphs, the bit at the bottom "risk of snow" with the pretty snowflakes, just how is it calculated? A percentage of the ensembles at that time who have precipitation at the location and a suitable 850s temp?

I believe it is based on the number of ensembles, happy to be corrected if wrong.

[Seasonality]

1 hour ago, nick sussex said:

To use a roulette analogy we're down to our last few quid, we've gambled nearly all the Christmas money and are now left with piling what we have left on one number!

Ordinarily at this point I would be throwing the towel in however theres just enough uncertainty both upstream and the exact placement of high pressure to the east to warrant keeping the towel until tonight.

The movement of the PV to the north is effected by what happens over in the USA, that's not to say its going to suddenly find some festive spirit and clear off! However its the modelling of the pattern over in the eastern USA which could either help or hinder our chances of getting some energy going se towards Iberia. Currently there are differences with the handling and amplitude of shortwave energy reaching the east coast, these have a knock on effect into the Atlantic.

I think current odds do favour a battle and then the PV winning out but don't discount yet that we might see more energy heading se.

Throwing in the towel on the evening of the 15th, that is pessimistic even for you Nick  With so much inter-run and inter-model variation and 10 days to go until the big day I think you should keep the towel a while longer. 

[Interitus]

38 minutes ago, warrenb said:

Well I apologise you want the train to run on the tracks.

No need for petulance, you could just download the MERRA2 stratosphere climatology to compare for yourself, for example -

The Glosea start date appears to be about 28/10/16 - that said some of the ensemble members don't appear to continue to the end which looks like multiple run times, but whatever.

From the MERRA data, the geometrically closest previous 16-day comparable periods (time of year for similar radiative history) of 10mb 60°N zonal wind to 28/10/16, were 17/10/09 and 24/10/04, let these be two 'ensemble members'. The similarity can be seen in the graph below leading up to the 'initialisation' day 31 marked by the vertical line, followed by the two ensemble members i.e. the wind speeds from 2009 and 2004, and their average the ensemble mean, alongside the current GEOS-5 analysis and forecast for comparison.

[phil nw.]

I think for the time being, with the way modeling has evolved, we have to steel ourselves for a period of increasing Atlantic influence as we go into next week.

ECM/GEFs height anomalies for T192 show the unsettled regime has reached further south and a glance at the London graphs indicate the uptick in rainfall around that time.

having said that there are suggestions from the later GEFs of the Greenland vortex weakening post Christmas with the draining away of those-ve ht anomalies to our north west-the trend starting to show around T300hrs.

The other change is the rebuilding of more +ve anomalies from the east so maybe week 3 will see the Atlantic repelled by a returning high.The interest will be in how this may manifest itself and whether we can see a higher latitude block develop.

 

 

Edited December 15, 2016 by phil nw.

[ArHu3]

42 minutes ago, Gael_Force said:

The only LRF output, of those we have access to, I have seen get the OND quarter nearly right is the Huug Van den Dool SSTA analogue forecast. The prediction back in August.

looks pretty accurate but I hope this model is off, it looks warm for the rest of winter and very cold for spring again...

[Bullseye]

1 hour ago, Gael_Force said:

The only LRF output, of those we have access to, I have seen get the OND quarter nearly right is the Huug Van den Dool SSTA analogue forecast. The prediction back in August.

The record warmth of November could be balanced out with the record cold of December over in the US, not often we one extreme to another in one month. This forecast for the US over OND could look fairly accurate come end of the month. Have they issued a forecast for the JFM period to see if any blocking sets up and whether its to our NW or NE or if the Euro high sets up once again? 

Edited December 15, 2016 by Bullseye

[ArHu3]

8 minutes ago, Bullseye said:

The record warmth of November could be balanced out with the record cold of December over in the US, not often we one extreme to another in one month. This forecast for the US over OND could look fairly accurate come end of the month. Have they issued a forecast for the JFM period to see if any blocking sets up and whether its to our NW or NE or if the Euro high sets up once again? 

 

(the image is in an opendir, you can browse it yourself and look through the entire forecast history )

Edited December 15, 2016 by ArHu3

[warrenb]

53 minutes ago, Interitus said:

No need for petulance, you could just download the MERRA2 stratosphere climatology to compare for yourself, for example -

The Glosea start date appears to be about 28/10/16 - that said some of the ensemble members don't appear to continue to the end which looks like multiple run times, but whatever.

From the MERRA data, the geometrically closest previous 16-day comparable periods (time of year for similar radiative history) of 10mb 60°N zonal wind to 28/10/16, were 17/10/09 and 24/10/04, let these be two 'ensemble members'. The similarity can be seen in the graph below leading up to the 'initialisation' day 31 marked by the vertical line, followed by the two ensemble members i.e. the wind speeds from 2009 and 2004, and their average the ensemble mean, alongside the current GEOS-5 analysis and forecast for comparison.

There was no petulance. Just merely pointing out that for a forecast it was pretty accurate with the trend, if you think a forecast is bad because is doesn't follow the exact actual data then you will never be very happy with any forecast

It is as if you are complaining that the day 10 ECM chart from 10 days ago doesn't match the reality.

When we get to that stage, there will be no need for forums such as these, because we will know the forecast models are correct.

 

It is as if you are arguing that Numerical modelling is pointless because we can use Analogs. If that is the case, then please show how the rest of the winter in your 2 analogs played out.

Edited December 15, 2016 by warrenb

[russell k]

1 hour ago, Ravelin said:

Since it's 'quiet time' between models runs I have a question. On the GFS ensembles graphs, the bit at the bottom "risk of snow" with the pretty snowflakes, just how is it calculated? A percentage of the ensembles at that time who have precipitation at the location and a suitable 850s temp?

Good question Ravelin, I rarely post on here, statistically it doesn't make sense to me. Is it computer generated or added on afterwards? There is also the version of the ENS showing a "snow row"

[SMU]

6 hours ago, fergieweather said:

Hi Steve: not sure which models you mean (missing the stratospheric trend), but GloSea5 certainly didn't. Here's snapshot of output from way back in late October (hence not latest modelling, which I can't show). Red is forecast mean; blue the hindcast mean. Note much weaker period foreseen during Nov; then marked uptick in 10hPa zonal winds (most members heading above average) during December. This signal was repeated and strengthened each run thereafter. UKMO were confident on the turnaround in SPV that has now unfolded, and GloSea5 timing for this was actually good. So it's incorrect to suggest UKMO missed this: quite the opposite; they were ahead of the game. 

Cheers Ian - A fair reflection of whats landed in reality is that forecast-

Whats clear from this & whats been clear perhaps for a couple of weeks- is that the 60N North tendancies have been modelled well thus far ( backed up by Dec 1-10 reanalysis V UKMO/GLOSEA  anomalies ) - however whats really hard & almost impossible to forecast is where the negative anomalies sit in respect to the core positive ones - & maybe something you would consider adding to your running commentary in terms of the forecasts going forward - 

I think the problem has been a 'little' misinterpretation of the blocking theme to the North / North East depicted from Late Nov ( despite your best efforts ) & people translating that into a final outcome of cold-

Obviously with a slightly sketchy picture there is always variations in terms of how people interprete what you write because in reality a block to the NE could mean cold for Greece or cold all the way across to the Uk- but rarely both..

With the negative anomalies added in the picture would be slightly clearer ! 

in terms of going forward this Winter, whilst you cannot show the GLOSEA outlook I would assume there will be a projected peak late Dec of around 50 M/S - almost on a par with 2015 then some moderation towards & into jan up to the end of Week 1

Thereafter the link with ERA interim data for the 00s ( more particular 2009 ) suggests a mid season SSW week 2 / 3 of Jan propergating for a reversal poss ~20th Jan

So a back loaded winter would be more the forecast now...

Best

s

Edited December 15, 2016 by Steve Murr

[Interitus]

15 minutes ago, warrenb said:

There was no petulance. Just merely pointing out that for a forecast it was pretty accurate with the trend, if you think a forecast is bad because is doesn't follow the exact actual data then you will never be very happy with any forecast

It is as if you are complaining that the day 10 ECM chart from 10 days ago doesn't match the reality.

When we get to that stage, there will be no need for forums such as these, because we will know the forecast models are correct.

 

It is as if you are arguing that Numerical modelling is pointless because we can use Analogs. If that is the case, then please show how the rest of the winter in your 2 analogs played out.

That is largely nonsense, but rather than being so defensive, how did the simple analogue example fare?

A good forecast should be an improvement over climatology, I have questioned that in this case and provided a reasonable example for why.

Edited December 15, 2016 by Interitus

[karlos1983]

ECM ensembles have got my attention towards the end, rather than massive spread it appears that we have a majority of milder solutions and a smaller group going for for cold. Interesting to see if the colder grouping grows in future runs. 

[Man With Beard]

1 hour ago, Gael_Force said:

The only LRF output, of those we have access to, I have seen get the OND quarter nearly right is the Huug Van den Dool SSTA analogue forecast. The prediction back in August.

I'm guessing that is pretty close to the mark. Well done that LRF!! 

[Nouska]

2 hours ago, Ravelin said:

Since it's 'quiet time' between models runs I have a question. On the GFS ensembles graphs, the bit at the bottom "risk of snow" with the pretty snowflakes, just how is it calculated? A percentage of the ensembles at that time who have precipitation at the location and a suitable 850s temp?

Why don't you pm @SylvainTV - he visits regularly and would clarify for you. I know the percentages are reflective of the ensemble probability section but not how they are derived so interested in the answer too.

http://www.meteociel.fr/modeles/gefs_cartes.php?ech=6&code=0&mode=26&carte=&proba=1

[bluearmy]

14 minutes ago, russell k said:

Good question Ravelin, I rarely post on here, statistically it doesn't make sense to me. Is it computer generated or added on afterwards? There is also the version of the ENS showing a "snow row"

I just assumed 2 members showing snowfall that day is 10% ??

[warrenb]

7 minutes ago, Interitus said:

That is nonsense, but rather than being so defensive, how did the simple analogue example fare?

Are we discussing at cross purposes, you used two analogues to show the Numerical GloSea was not so good, I asked wha those 2 analogues showed for the rest of their respective winters.

[russell k]

3 minutes ago, bluearmy said:

I just assumed 2 members showing snowfall that day is 10% ??

Thanks

So did I, but is the actual chance of snowfall for that location actually 10% on that day? or if it showed 5 different days at 10% I would have a 50% of snow in that time frame? so on up to 100%?

[Glacier Point]

500 hPa height anomaly for the NH for the first 12 days of the month not really reflecting the likely outcome for the month's end and certainly not the expectation from earlier on. 

The core tropical forcing signature is moving back to where one would expect given sub seasonal drivers - centred to the Western Pacific. Contrary to what be expected looking at some model outputs, zonal wind anomalies remain modest and towards the weak category across much of the mid to high latitudes all the way up to 30hPa. The presence of a sub-tropical jet flow very evident looking at sectional plots.

And yet, a pattern forecast that refuses to reflect these and other sub seasonal drivers. Why ?

These are all likely related, but I suspect that three things are relevant here.

1. Drop in angular momentum with southward momentum transport across 30N - GEFS projection of a -1.5SD reduction look as if they'll verify.

2. West QBO phase which is on steroids, December's index value will trump all other Decembers by a mile.

3. Arctic sea ice registering at nearly 1.5m sq k below the climatological norm.

I think that latter might be the most persuasive given this is the factor which defines this year in difference to others given similar forcing. Does the Arctic have to bottle up cold in order to restore some kind of equilibrium to surface temperatures there ? 

Looking ahead, I get the model consensus to develop a substantive Euro / Scandinavian ridge. That tallies with what I expect to be a dominant subseasonal player orchestrated by the arrangement of SSTAs across the Western Pacific. The inference for January being a withdrawn Bermuda ridge and trough angled NW-SE across the eastern part of the North Atlantic. This composite probably needs a tweak or two to reflect increase heights over Scandinavia, but the broad gist looks about right.

With a large corrective drop in angular momentum and assuming no substantive upward corrections (can't rule that out), GWO likely to be low amplitude phase 2 for the remainder of December into week 2 January. Composites for these on a monthly basis:

>>

Again, not counter to the Euro / Scandinavian ridge, if anything whiff of the ridge extending westwards but I'd caution interpretation of surface temps on that basis. Note the suggestion there of continued polar vortex lobe west of Greenland, sufficiently far west to allow for the Scandinavian/ Euro ridge to breathe. That composite for December probably being swamped by factors outlined above, so direct interpretation not advised, but an attempt to build a second ridge in the North Atlantic merging with the ridge to the east of that would be something to look for, albeit in a likely suppressed form.

One other thing to look for, and perhaps of greatest interest, will be pattern projected for at least the next 10-15 days. +ve height anomalies over Eastern Europe / Western Russia will be geo-effective for Wave 2 increase in the low to mid stratosphere, taking effect early January.

GFS typically beginning to reflect this a little prematurely but I would expect to see splits developing in the trop vortex within the 1st week of January. The combination of cold air potentially entering the fray far stage right and +ve height anomalies over Scandinavia ? One for those of a nervous disposition to look away. Can we lower heights sufficiently to our south over Iberia to advect this cold pool westwards ?

In the meantime, dry and relatively mild conditions look the form horse until perhaps some interest in the modelling last week of the month.

Edit: just noticed. Compare GEFS plot above with composites for tropical forcing centred over the West Pacific / Maritimes - good reflection across the North Pacific, North Atlantic and Scandinavia. Seems to be capturing much of the NH forecast pattern end December.

Edited December 15, 2016 by Glacier Point

[Interitus]

6 minutes ago, warrenb said:

Are we discussing at cross purposes, you used two analogues to show the Numerical GloSea was not so good, I asked wha those 2 analogues showed for the rest of their respective winters.

Not cross purposes, I compared climatology favourably with Glosea and hoped for your opinion as part of reasoned discussion.

So meteorological agencies can run multiple updates of numerous members with fresh data throughout the season, but I can only use two analogues from October? Seems fair. As it happened by coincidence both analogues reached peak wind speed exactly 85 days from start date. After this they had SSWs 23 days apart - something of a divergence but as would a random pair of ensemble members. We're not allowed to see these, but fortunately, the climatology is here - https://acd-ext.gsfc.nasa.gov/Data_services/met/ann_data.html

No more cluttering this thread from me.

[bluearmy]

The image for the first 12 days of the month would have been excellent for coldies if the Euro heights had subsided

as in many a disappointing coldies  winter for nw Europe, that remains the overriding feature imo 

[bluearmy]

24 minutes ago, russell k said:

Thanks

So did I, but is the actual chance of snowfall for that location actually 10% on that day? or if it showed 5 different days at 10% I would have a 50% of snow in that time frame? so on up to 100%?

Each member is a different percentage chance - 5%

so you would have a 10% chance of snow on each day. Can't see how you could accumulate your overall chance to be 50% over five days. You could say that you have a 50% chance of seeing snow on any given day within that five day period - I think but that would assume that each day a different member was predicting snowfall .  If the same member predicts snowfall for five days then surely that remains a 5% chance of seeing snowfall at some point In the five days rather than a 25% chance of seeing snow at some point in the five days

i think !

 

[blizzard81]

5 minutes ago, bluearmy said:

The image for the first 12 days of the month would have been excellent for coldies if the Euro heights had subsided

as in many a disappointing coldies  winter for nw Europe, that remains the overriding feature imo 

Indeed. The question is, what is stopping the lowering of heights over southern Europe winter after winter?

[Matthew.]

7 minutes ago, blizzard81 said:

Indeed. The question is, what is stopping the lowering of heights over southern Europe winter after winter?

Could it be CC? Since 1993 warmer winters really prevail on the whole. Southern height anomalies might be linked to this in a warming world. Just a thought I think I read somewhere.

[Singularity]

4 minutes ago, Matthew Wilson said:

Could it be CC? Since 1993 warmer winters really prevail on the whole. Southern height anomalies might be linked to this in a warming world. Just a thought I think I read somewhere.

A net poleward expansion of the Hadley cell may be a manifestation of climate change to blame (I'm not sure how compelling the evidence from research is for this?); that means stronger Azores Highs which can ridge further into Europe and with greater frequency. I think the greater equator-pole difference in winter then leads to a stronger effect than seen in summer, much to our discontent.

All very theoretical though... and off topic so I'll stop there.

Regarding @Glacier Point's intriguing update, GEFS has been forced to delay the marked GLAAM drop by about a week already, and the latest update has even introduced a slight positive tendency for tomorrow, so I'm far from convinced as to such a large drop taking place. So to GP I ask, what are you aware of that makes it look like it will verify?

Also, I'm not sure how the low sea ice could force a bottling up of tropospheric cold air given the reduced equator-pole temp gradients allowing more jet meandering where and whenever the tropical or stratospheric forcing is not directing it to do otherwise - indeed there's been a lot of spilling out of cold air to the low latitudes, and import of warm air from them, in recent times - hence the record low ice extent being in place to begin with. While storms may become more intense in the Arctic basin, such strong systems also tend to draw in relatively warm air from the lower latitudes. If you could enlighten me - and others on here - as to what's behind that hypothesis, that would be much appreciated.

TIA