Model Output Discussion - Spring Has Sprung

[KTtom]

ECM backing the trend of an Atlantic trough in the run up to Easter..cant see anything out there that hints of a warm dry easter weekend, still time for a change but the trend isnt wonderful.

[Bats32]

 KTtom Been watching the models for a while as I go on a cruise on Easter day and that trough just seems so locked on.

[Cambrian]

Here at day 5 on the 12z GEFS, as we’ve seen several times during the last few months, the hemispheric profile achieves a striking north-south orientation of heights, the PV with its core over Siberia coordinating a Scandinavian trough,

 

which combined with cross-polar heights at the merging of Alaskan and Atlantic ridges, delivers us a handsome northerly flow.

But as during the winter, it’s brief. The pattern has no hold. By day 10, the PV relaxes, the Siberian core leaks out west again towards northern Greenland, so the pattern falls lateral again and the reinforcement of the Greenland trough drops a low pressure southeast out of Greenland. 

 

The Alaskan heights are squeezed out, and there is a complete reversal over the Atlantic - with a 40+ dam drop in heights over 5 days to the west of the UK and Ireland, we’ve swiftly lost the Atlantic ridge and we’re back into an Atlantic cyclonic pattern heading into Easter. 

And as during the winter, once it’s set up, this pattern is not brief. By day 15, there are heights to the north but the PV is lateral, and still very much on our side of the pole, preventing a full build, so instead, we get a southerly tracking jet stream brewing up systems carrying a lot of moisture running up from the southwest, through the corridor paradoxically being held in place by the northern heights.

 

It’s as we were then, seemingly too far south for the northern heights to bring us blocked cold, and too far north for the southern heights to send us a proper block, so we get a fleeting Atlantic ridge that can’t form a robust block with the heights to the north, because they themselves lack sufficient traction, followed by the PV and shallow heights to their south then going lateral and serving us up another extended run off the Atlantic.

Fair enough to point out that is what we would expect for a temperate Atlantic climate but the unrelenting nature of never more than a few days with any other option on the table, at the moment at least, seems quite grimly exceptional. 

[damianslaw]

 Cambrian Good post. This scenario has by and large played out since July, brief drier incursions, the odd very brief northerly, but by and large an omnipresent deep trough has sat either to our west, or SW, pulling in either mild southerly sourced air or more often than not dank south westerly/westerly airstreams, with the north at times tapping into a cooler source as the jet sinks south.

Still a long way off, but the models continue to show a rather unsettled outlook in the run up to easter after a chilly northerly this weekend. 

 

 

[ANYWEATHER]

Startling differences between ecm and gfs at day ten......

[Froze were the Days]

ECM going the route of 'the only way is the Norwegian Sea' and the effects of heights over the near continent just too much for any low pressure (saying that has been the way for some time)...my money is on the ECM being closer to the mark and the first run of late GFS has gone to sustain heights to the north from the shorth northerly.

Roll on Spring I say...who wants cold now?

[Rain All Night]

10 hours ago, Rain All Night said:

0z deterministic runs, out to Mon 25th (day 7)

The northerly looks quite nippy for this weekend.

0z ensemble means, out to Tue 2nd (day 15)

The signal is unfortunately clear and consistent for a "big trough" arriving later next week...

12z deterministic runs, out to Mon 25th (day 7)

The "wedgyness" shown by some of the models this afternoon - particularly GFS (which I can see as of writing this is going the same way on its 18z run) - is a curiosity, but judging from the ensemble means seen below, it looks doomed to come to nothing.

12z ensemble means, out to Tue 2nd (day 15)

No change to the prognosis for the last week of March - just look at that trough!

[andymusic]

 Froze were the Days gfs 18z really trying it's best for the ski industry in Scotland with cold from this weekend until the beginning of April on this run

[KTtom]

 Bats32  At the timeframe we are looking at you would expect some wild variations thrown up by the ops but like you say GFS been pretty unrelenting in the Atlantic trough dominating Easter weekend. 00z mean below a carbon copy of yesterdays ECM I posted.

[MJB]

Much colder next mid week from GFS 

 

 

Quite different from ECM

UKMO in 7 days

Looks a miserable Easter 

Edited March 19 by MJB

[ANYWEATHER]

Why people are writing Easter off and it's over ten days away is beyond me when model accuracy is five day maximum. The charts I've posted this morning for Good Friday will change drastically and look nothing like the charts this morning come Good Friday.

[Alderc 2.0]

 ANYWEATHER I don’t think anyone is writing Easter but it’s an unrelenting pattern of Atlantic dominance being shown. The only good thing I can see is it looks like there is not going to be a repeat of the spring drought in southern Europe like last year. 

[ANYWEATHER]

 Alderc 2.0 it’s not the fact that it may or not happen , just the fact that some members are referring to 10days plus away as reliable……

[johnholmes]

Too far to be sure what Easter will bring but the ECMWF is a fairly good indicator of both upper air and surface along with 850 temperatures. It only goes out to Maundy Thursday but not a bad idea to follow each subsequent update from now

And of course the 500 mb NOAA Anomaly charts the 6-10 out to the same day and the 8-14 over the Easter weekend. This just for an idea of where the main upper air features look like being. 

It does, at the moment look like an upper trough will be close to the UK with the 500 flow coming from just north of west.

Climate Prediction Center - 8-14 Day 500mb Outlook

WWW.CPC.NCEP.NOAA.GOV

Time will tell, as always, the accuracy of the above charts

[Summer8906]

I simply can't believe our endless bad luck. 9 months of endless dull, wet gloom and now the models are showing one of the worst Easters on record.

[ANYWEATHER]

16 minutes ago, Summer8906 said:

I simply can't believe our endless bad luck. 9 months of endless dull, wet gloom and now the models are showing one of the worst Easters on record.

How can you make a post like this when we are ten days + to Easter?? Bit premature to be writing off Easter when there is no reliable weather data  at the moment to verify what the weather will actually be...!☺

[Summer8906]

 ANYWEATHER Just seems to be showing again and again on most model runs, I guess.

[johnholmes]

 Summer8906

So where in Hampshire do you live. I'm sure the weather stats for Bournemouth are nothing like as bad as your post suggests?

[Rain All Night]

14 hours ago, Rain All Night said:

12z deterministic runs, out to Mon 25th (day 7)

The "wedgyness" shown by some of the models this afternoon - particularly GFS (which I can see as of writing this is going the same way on its 18z run) - is a curiosity, but judging from the ensemble means seen below, it looks doomed to come to nothing.

12z ensemble means, out to Tue 2nd (day 15)

No change to the prognosis for the last week of March - just look at that trough!

0z deterministic runs, out to Tue 26th (day 7)

GFS and ICON are still finding height rises to mix things up a little at the beginning of next week.

0z ensemble means, out to Wed 3rd (day 15)

The one bit of hope to cling to is that the trough signal seems to peak just before the start of the actual Easter weekend. The GEFS is as keen as the GFS on the idea of "wedgyness" after the northerly this weekend, perhaps slowing the trough down a bit.

Edited March 19 by Rain All Night

[KTtom]

 ANYWEATHER Who has written off Easter?  The discussion is the constant there of an Atlantic trough to the west in the build up to the weekend as shown by the models and being discussed in the model thread. This forum would be pretty boring if we could only discuss model output for 1 day ahead!

[ANYWEATHER]

 KTtom at the same time the mad thread gets a bit monotonous when discussing weeks in advance which we know won’t happen

[KTtom]

 ANYWEATHER totally agree when someone posts a rogue BFTE when the last 15 runs have all being mild but, I say again, the discussion you seem to have taken exception to was a trend of an overall pattern which has been a regular theme over the last few days, agree, detail will change but the overall pattern being shown does match the Mets forecast.

[Alderc 2.0]

 johnholmes  Well actually they are, nearly 100mm of rain and only about 40hrs of sun. Off the back of the dull and wet February it’s been really grim for the last 7 wks. 

[Cambrian]

There’s often comment on here regarding the reduced reliability of the models past a given timeframe. 5 days is often stated as the point at which models become unreliable.

Studying the ECMWF web pages, it would appear to be a good deal more nuanced than this.

They publish charts showing the “Continuous ranked probability skill scores (CRPSS) of forecasts of upper-air parameters by TIGGE centres”, with the following notes about the score being used:

The Continuous Ranked Probability Skill Score (CRPSS) is a measure of how good forecasts are in matching observed outcomes.   Where:

CRPSS = 1 the forecast has perfect skill compared to climatology - forecast beneficial;

CRPSS = 0 the forecast has no skill compared to climatology) - forecast has no benefit over climatology;

CRPSS = a negative value the forecast is less accurate than climatology - forecast misleading.

CRPSS is evaluated by calculating the function  CRPSS = 1 − CRPSforecast / CRPSclimat where:

Continuous Ranked Probability Score for the forecast (CRPSforecast) is calculated comparing the Cumulative Distribution Functions (CDF) for the forecast against observations (or analyses) over a given period. 

Continuous Ranked Probability Score for climatology (CRPSclimat) is calculated comparing the Cumulative Distribution Functions (CDF) for the forecast against climatology over the same period. 

Hersbach, H., 2000: Decomposition of the continuous ranked probability score for ensemble prediction systems. Wea. Forecasting, 15, 559-570.

Here’s the link to the paper that ECMWF cite:

https://journals.ametsoc.org/view/journals/wefo/15/5/1520-0434_2000_015_0559_dotcrp_2_0_co_2.xml

From the cited paper, excerpted: Instead of two options (event occurs or does not occur), the range of the parameter of interest is divided into more classes. In addition, the RPS contains a sense of distance of how far the forecast was found from reality. For a deterministic forecast for instance, the RPS is proportional to the number of classes by which the forecast missed the verifying analysis. Although the choice and number of classes may be prescribed by the specific application, the exact value of RPS will depend on this choice. It is possible to take the limit of an infinite number of classes, each with zero width. This leads to the concept of the continuous ranked probability score (CRPS)…This CRPS has several appealing properties. First of all, it is sensitive to the entire permissible range of the parameter of interest. Second, its definition does not require…the introduction of a number of predefined classes, on which results may depend….Finally, for a deterministic forecast, the CRPS is equal to the mean absolute error (MAE) and, therefore, has a clear interpretation.

 

This is useful because:

1. Rather than the (subjective) right or wrong, we have a graded measure of the extent of deviation from the realised outcome. 

2. Any score above zero demonstrates some forecast skill over the assumption of climatology. 

 

Here are the CRPSS charts for T850 and 500hPa height for the Northern Hemisphere incorporating data for the winter just ended. At day 5, ECM attains a score of 0.62 for T850 and a score of 0.72 for 500hPa geopotential. The score for heights is higher at day 6 than that for T850s at day 5. 

 

The timeframe above the 0.5 CRPSS level (where the forecast matches the observed outcome in more instances than it doesn’t, compared to climatology) corresponds to between day 6 and 7 for T850 and between day 7 and 8 for heights. 

At day 10, the score is 0.28 for T850 and 0.33 for 500hPa height. Both those scores continue to offer reasonable “value” at day 10, though to be fair, at a level less than half the value offered at day 5.

By day 15, the score is around 0.12 for T850 and near 0.15 for 500hPa height, much lower, but still above zero so exhibiting some skill when compared to just assuming climatological norms.

The ECMWF appears to put some store in the 0.25 or 25% level, as this is what they use to illustrate the improvement in skill over the last 25 years. This graph is great.

It shows the lead time in days before the score drops to 25% and is for T850. The model has the same level of skill at day 10 in 2024 than it did at 5.5 days in 1998. Noticeable too that winter T850s appear to be handled better than summer ones, probably due to the winter PV, with typically fewer meridional episodes and less dramatic fluctuations in the winter as a result. 

In summary, in our parlance, the models are more “reliable” for longer for predicting heights than they are T850s. For heights, these are typically of sound merit all the way out towards forecast day 8. 

The reliability drops off steadily from day 0, there is no sudden drop at any point and the forecasts maintain some predictive value when compared with climatological expectations all the way through to day 15, which is presumably why this timeframe is catered for in the outputs. 

Lastly, and perhaps most excitingly, the models appear to be better now at day 10 than they were 25 years ago at day 6. 

This all fits very nicely with the general narrative surrounding their use here in the model thread. 

[*Stormforce~beka*]

 johnholmes Bournemouth is in Dorset which is sunnier than Hampshire. Also on the Coast inland Hampshire has had a horrific time