Jump to content
Sign in to follow this  
Captain Shortwave

Model Output Discussion - The final stretch to Christmas

Recommended Posts

Isn't this really disappointing in that the higher resolution might end up causing more problems than it solves?

 

In an effort to improve the model in a sense they've made it worse!

 

Agree.  Another model adds to the confusion.  The only way of resolving this is to run  regular verification tests over a long period of time to assess the variances in bias and reliability.  So far I am not convinced.  Too many cooks spoil the broth.

  • Like 1

Share this post


Link to post
Share on other sites

Good point ba.Would you consider it has a lesser reliability factor than the gfs op then?Sm seems to rate it but im not sure why tbh.thanks

It seems to me that the GFS P has been giving us the most cold outcome in FI out of all the models for quite some time but they never materialise. I end up checking it for fun rather than taking it seriously. No wonder they keep postponing its start date.

  • Like 4

Share this post


Link to post
Share on other sites

the stats ive seen of the gfsp seem to place it third, just above its brother. given the amount spent, i suggest NOAA wont be too pleased!  early days though and it may well turn out to have its plusses in certain evolutions. i would have thought it should do a better job in picking up an end of week 2 trend, given that it runs to day 10 in high res.  forgetting the detail, it will have a handle of sorts on the hemispheric longwave pattern at day 10.

 

and i'm not keen on the tooing and froing of the vortex. i would rather it stayed in hudsons or nw siberia. its movement tends to fire up the northern arm. there does appear to be space for a svaalbard/scandi height rise in week 2 as the hudsons chunk takes over. if the subsequent northern arm push can throw a big enough ridge ahead of it and split to our west, we could be looking at a good block for coldies. otherwise, its the zonal train with euro ridge.

  • Like 2

Share this post


Link to post
Share on other sites

Yeah, what Tamara said. :unsure2:

 

 

Re: Verification. Assuming the parallel is the PRX (although I believe there's some debate over whether it is a slightly different version), then it is actually lagging its older brother.

 

cor_day5_HGT_P500_G2NHX.png

  • Like 3

Share this post


Link to post
Share on other sites

Can you provide any evidence that they create LRFs based on isolated mean charts? No interpretation can be drawn from them, only that, the further they get from zero hours, the more worthless they are.

 

See below.

 

 

 

 

I know CFS and JMA monthly forecasts are based on the means (or the anomaly of). UKMO (via MOGREPS) are ensemble based (15 days/22 member).  EUROSIP is like the UKMO Winter forecast and is based on probabilities and anyway EUROSIP is a multi model system. The ECMWF have various outputs on their LR suite but one of the main ones is: 

"Climagrams

Time-series of monthly mean anomalies. Available parameters are: area averages of sea-surface temperature2m temperature,  precipitationteleconnection and monsoon indices."

 
source:  here
 
So yes the mean is used in conjunction with other data for the pros. What we, the general public get to see in the LR is usually the mean charts. :(
  • Like 2

Share this post


Link to post
Share on other sites

 

I know CFS and JMA monthly forecasts are based on the means (or the anomaly of). UKMO (via MOGREPS) are ensemble based (15 days/22 member).  EUROSIP is like the UKMO Winter forecast and is based on probabilities and anyway EUROSIP is a multi model system. The ECMWF have various outputs on their LR suite but one of the main ones is: 

"Climagrams

Time-series of monthly mean anomalies. Available parameters are: area averages of sea-surface temperature2m temperature,  precipitationteleconnection and monsoon indices."

 
source:  here
 
So yes the mean is used in conjunction with other data for the pros. What we, the general public get to see in the LR is usually the mean charts. :(

 

All these models produce mean charts IDO but the pros dont use the mean charts from these models to produce a forecast. They study in FAR greater detail than that the individual runs and clusters. Then what happens is they look at the background signals that Tamara has just spoken about ^^^^^^^^^ and work out which of the clusters is the most likely. Don't get dragged in to thinking the means are as important as you think and most importantly dont post these to make novices think what they show is the most likely solution

  • Like 5

Share this post


Link to post
Share on other sites

 

 

At least much of this interest remains favourable for those whose hopes are cold hearted :)

Great read. So when do you think this cold pattern will take hold? It seems the disconnect is causing a continual headache.

 

This hyped winter could end up forgettable due to simple disconnect with regards to the Nino.

  • Like 3

Share this post


Link to post
Share on other sites

All these models produce mean charts IDO but the pros dont use the mean charts from these models to produce a forecast. They study in FAR greater detail than that the individual runs and clusters. Then what happens is they look at the background signals that Tamara has just spoken about ^^^^^^^^^ and work out which of the clusters is the most likely. Don't get dragged in to thinking the means are as important as you think and most importantly dont post these to make novices think what they show is the most likely solution

 

I agree the Pros are paid the big bucks to use their training and experience to break down the mean and clusters to try and get us some sort of accurate forecast. I was however talking about the computer models. They produce a suite of products and the main system are the means, the foundation for many LRF. These computers cost tens of millions to come up with their LR means/clusters, so its a bit harsh to say they are wasting their time :doh:

 

Lets hope the GFS P is right and the current horrible mean can be discarded :)

Share this post


Link to post
Share on other sites

Yeah, what Tamara said. :unsure2:

 

 

Re: Verification. Assuming the parallel is the PRX (although I believe there's some debate over whether it is a slightly different version), then it is actually lagging its older brother.

 

cor_day5_HGT_P500_G2NHX.png

 

Although at day 10 it is performing better (which you would expect, given the focus on improving that area).

 

cor_day10_HGT_P500_G2NHX.png

Share this post


Link to post
Share on other sites

Yeah, what Tamara said. :unsure2:

 

 

Re: Verification. Assuming the parallel is the PRX (although I believe there's some debate over whether it is a slightly different version), then it is actually lagging its older brother.

 

cor_day5_HGT_P500_G2NHX.png

 

In fairness though, 0.005 is not much of a lag. :D

Edited by radiohead

Share this post


Link to post
Share on other sites

 

I know CFS and JMA monthly forecasts are based on the means (or the anomaly of). UKMO (via MOGREPS) are ensemble based (15 days/22 member).  EUROSIP is like the UKMO Winter forecast and is based on probabilities and anyway EUROSIP is a multi model system. The ECMWF have various outputs on their LR suite but one of the main ones is: 

"Climagrams

Time-series of monthly mean anomalies. Available parameters are: area averages of sea-surface temperature2m temperature,  precipitationteleconnection and monsoon indices."

 
source:  here
 
So yes the mean is used in conjunction with other data for the pros. What we, the general public get to see in the LR is usually the mean charts. :(

 

 

CFA and JMA are at a very extended range, we're referring to the range that the main models cover.We can see the GEFS ensembles which come out at the main time and the De Bilt ensembles are of some use but just because we can see a mean it doesn't mean that it illuminates anything on its own. The key words are 'in conjunction'.

Share this post


Link to post
Share on other sites

I agree the Pros are paid the big bucks to use their training and experience to break down the mean and clusters to try and get us some sort of accurate forecast. I was however talking about the computer models. They produce a suite of products and the main system are the means, the foundation for many LRF. These computers cost tens of millions to come up with their LR means/clusters, so its a bit harsh to say they are wasting their time :doh:

 

Lets hope the GFS P is right and the current horrible mean can be discarded :)

Yeah you are right mate theyre far better than us at reading between the forecasting lines. Im just saying as a bit of advice from experience and knowing how the met work long-term not to post long term charts and post lines that can come across a bit sweeping. The language used was as though because of the mean charts zonality was inevitable. It can be mis-leading. It really is not inevitable as the models dont know whats happening past 120 hours. Id concentrate no further than that, honestly. Read Tamaras excellent post above^^^^ for long term trends. Zonality isnt the form horse based on back ground signals. Youd expect models to slowly trend to a more blocked scenario, I think a poor signal was picked up yesterday.

 

Amen to that GFS P!!! Merry christmas to that my friend!

  • Like 1

Share this post


Link to post
Share on other sites

At the risk of being crude, uneducated, and generally out of my depth with the science, my broad impression of things from closely following the model outputs since the start of november, and leaving aside for the moment the (hugely insightful and educated) contributions and discussions, is that the consistent pattern has been resulting in a majority outcome of mid-lat anticyclonic blocking.

 

This 'middle-ground' outcome seems clear, in that no dominant pattern of more extremely defined synoptics has emerged, during this long period. There has been a very gradual period of overall cooling in this region, almost simply an expression of 'normal' seasonal shift. This cooling seems for the moment to have ceased, perhaps (no doubt naively!) in response to the fact that we are at the solar minimum.

 

This sounds like stating the obvious but I state it (having rarely posted) for the reason that it's often frsutrating and baffling to read some or many of the forecasts interpreting much more 'polarised' (pun intended) patterns of weather from what seems like a trend that is consistently 'middle of the road'. There has not been enough evidence from all of the outputs that I have seen to suggest any further move towards anything much different.

 

Perhaps the gradual shifting of things on the more global/hemispheric scale will continue to produce gradual shifts in our synoptics in this region but whether it will produce any appreciable shift in our weather overall seems on the evidence of the past few weeks unlikely.

 

Again to take the crude view, in an effort to escape the mire of multiple information and in-depth interpretation, the hemisphere has not generally looked especially 'cold' for the past month. Admittedly November brought a severe lobe of cold to NA, and the potential for that to happen again, and to shift eastwards, appears decent enough. But there remains alot of 'mild air' over our hemisphere as far as I can see, and alot of persistant lower to mid-lat heights, and the thing I'd love to know is how this might shift away?

 

There has been no pattern to penetrate very cold Arctic air far enough south. Or to move the mild anticyclonic bands away, except temporarily. This kind of pattern in my memory tends to last up to 3 months at times in winter. That crude and narrow impression leads me to believe it's not going to shift much in the next 2 weeks. Everything I have seen over the past week from the forecasts has indicated to me that mostly quiet weather will be the majority, with mostly average temperatures, and occasionally more unsettled spells.

  • Like 3

Share this post


Link to post
Share on other sites

In fairness though, 0.005 is not much of a lag. :D

 

If you look at the 'die off' charts as below for nearly all of the parameters, you can see ECM still outperforming all up to 192h, with the GFS and parallel similar but then a dramatic improvement between 192h and 240h. I should think that they are still scratching their heads as to how to approach ECM before 192h.

 

http://www.emc.ncep.noaa.gov/gmb/STATS_vsdb/

Share this post


Link to post
Share on other sites

In terms of " Meangate" !

 

There are times when this helps if for example you're dealing with straightforward patterns say for example northerly topplers at day ten.

 

Mean values are however useless at undercutting scenarios where generally the ensembles might want to take too much energy ne rather than se, the difference in outcome is huge for temps in the winter months.

 

The clusters are IMO better because with background signals you might be able to edge towards a particular cluster.

 

Todays ECM mean exhibits two distinct clusters one with low pressure to the nw and one with low pressure further se, both are the result of either energy split or phased energy to the nw. The mean shown at day 6 from the ECM dilutes the actual by sticking the low over the UK when in fact low pressure will either be to the nw or se.

 

You don't always have these types of large divergences at relatively early timeframes for this reason the mean sometimes needs to take a backseat, you have to use what best fits a certain situation.

Thats summed up excellently Nick. The 28th onwards is extremely interesting in my eyes

  • Like 3

Share this post


Link to post
Share on other sites

Thats summed up excellently Nick. The 28th onwards is extremely interesting in my eyes

Thanks. Just to highlight the divergence the wind directions for the ECM ensembles:

 

post-1206-0-91840900-1419261945_thumb.pn

 

As you said the 28th is looking interesting and IMO the pivotal moment, let the nerve shredding begin!lol

Edited by nick sussex
  • Like 3

Share this post


Link to post
Share on other sites

Just caught up with another day of many pages.... LOVE the fast paced forum in winter. Exciting times ahead :D  Some great posts to read today so far!

 

Its time for the 12zs so heres hoping the cold rebound continues from this morning........

Share this post


Link to post
Share on other sites

If you look at the 'die off' charts as below for nearly all of the parameters, you can see ECM still outperforming all up to 192h, with the GFS and parallel similar but then a dramatic improvement between 192h and 240h. I should think that they are still scratching their heads as to how to approach ECM before 192h.

 

http://www.emc.ncep.noaa.gov/gmb/STATS_vsdb/

At 500hpa, the higher res of the para post day 8 looks to be paying dividends wrt the normal Gfs.

Share this post


Link to post
Share on other sites

 

 

Yesterday was a good time to stand back and not post anything. The unreliability of the modelling and pendulum swing attempt to ignore the background atmosphere drivers seemed obvious. The manifestation of this misread of signals was the apparent flip to much more consensus for flattening out the pattern over progressively to a zonal one, and should not have seemed convincing in the circumstances. Any further such modelling, in prevailing circumstances, should continue to be viewed with suspicion - until, or. if those circumstances change...

 

 

 

You say that the models are misreading the signals and ignoring the background atmosphere drivers, but what exactly do you mean there?? The models if anything have these factored/built into them and produce there runs with everything that's going on factored in.  

 

The reason we are having such wide discrepancies run to run is exactly because there are NO CLEAR signals of what will happen and as models churn out there runs pretty much through mass number crunching, small differences at the start are obviously going to have big implications later on (the butterfly affect if you will.)

 

So really ''Any further such modelling, in prevailing circumstances, should continue to be viewed with suspicion - until, or. if those circumstances change...''  doesn't make sence as we are always going to be suspicious about each and every run as all it is is basically one possibilty been shown at any time and with no guarantees of it been correct.

 

And regards to a pendulum swing attempt to ignore the background atmosphere drivers, the models don't ignore anything, like I've said they just go on the same data that is ALWAYS fed into them and go from there, they certainly don't ignore anything lol   :unsure2:

Edited by Harsh Climate
  • Like 7

Share this post


Link to post
Share on other sites

We have all experienced model disagreements before (where one will show cold/mild and the others the opposite) but I have never known all the models to flip between a whole host of solutions almost daily.

 

To me this smacks of something so unusual going on that the models have not been programmed to respond to. Let's hope it is a set of circumstances that result in a prolonged cold spell.

  • Like 5

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.
Sign in to follow this  

×
×
  • Create New...