Model Output Discussion - Summer 2024

[WYorksWeather]

GFS 12z tonight proves my point about the importance of detail in terms of the wind direction at this time of year.

At day 5, just a slight tilt of the high.

At day 6, you can almost trace the wind flow arrows back - wind being sourced from slightly further south, perhaps more towards the Balkan states.

The overall impact is pretty obvious when you then look at the temperature charts. All of next week considerably warmer than average and dry if this forecast is right.

To a lesser extent this also continues into the following week on this run, but I didn't show that since it's definitely FI by that point.

In terms of the ensemble, the GFS is perhaps fractionally above its mean, but not extraordinarily so.

It is this sort of detail that will determine how next week goes - source the air just from slightly ESE makes a big difference compared to E or ENE at this time of year.

 

[damianslaw]

The models are showing quite an unusual profile for the time of year, much blocking and no energy in the jetstream. Tropical storm activity also very low.

Where is the jetstream? More akin to April and May than September.

The new week set very fair indeed, pleasant with an easterly airflow. Why oh why could these synoptics not verify April- August!

[raz.org.rain]

 damianslaw winter synoptics in summer, summer synoptics in winter. 2024 will likely come out as the year that satisfied nobody.

[Don]

 WYorksWeather The GFS has dropped it's retrograde signal in FI, but has been known to spot a trend in the past, drop it and then pick it up again a few days later.

[stodge]

 damianslaw I think I remarked on this a couple of weeks ago.

It seems related to the northward shift of the ITCZ which has caused the Sahara rainfall and meant the waves exiting Africa have not developed into tropical depressions/storms despite the conducive environment in terms of SSTs in the tropical central Atlantic and of course the Gulf of Mexico (as we've seen this week with Francine, it only needs a short period of favourable conditions to move a relatively innocuous depression into a Category 2 event).

Why the ITCZ has moved north I don't know - will this mean a "later" hurricane season with storms in October and even November -this may relate to forecasts of a wetter than usual period for the British Isles if the usual strengthening of a jet is re-enforced further by ex-tropical energy? Put a strong ex-hurricane and its energy into the jet and see what happens....

The other possibility is the jet won't strengthen to the usual extent and we'll have a more blocked winter - I don't see it myself but this summer has confounded a lot of experts and their LRFs. I was in North Norfolk earlier this week and it definitely felt autumnal yesterday and the day before.

[WYorksWeather]

 damianslaw Very weak polar vortex is probably one factor. It's really struggling to get going at the moment.

However, the ECM forecast does show it strengthening rapidly and reaching climatological strength by mid-October, and then becoming stronger than average by the end of the month. Of course this is a long way off and the zonal wind forecasts performed poorly last year. However, if it were to come off, it would suggest increased likelihood of blocking (which of course can be warm or cold in September and October) for the next few weeks, but then perhaps a return to a more typical autumnal pattern for the second half of the autumn?

[Daydream Boy]

Could we possibly change the thread title to "Heading into Autumn"?

[WYorksWeather]

EC clusters below.

Days 3-4

Easterly setting up on Monday and Tuesday - no surprises here.

Days 5-7

Cluster 1 (73%) has an easterly from day 5 to day 7. Should be generally warm and dry.

Cluster 2 (27%) has slightly more of a tilt to the high. Very similar to Cluster 1, but looking at the finer detail perhaps more of an ESE wind direction rather than a flat easterly. Hence likely to be warmer, perhaps a couple of very warm days in favoured spots. 

The OP was in Cluster 2.

Days 8-10

Down to just the one cluster here, so I would be careful about over-interpreting this as it will mainly reflect the ensemble mean.

Remaining generally warm at day 8 and 9 with easterlies. By day 10, hints of the high pulling to the NW, so a cooler north-easterly.

Needless to say, the OP was in Cluster 1!

Days 11-15

A lot of options here. Usual warning about uncertainty in the extended range beyond day 10.

Cluster 1 (25%) retrogresses the high somewhat allowing a cooler flow from the NW at day 11 and day 13, then flattening off to a fairly standard unsettled westerly pattern by day 15.

Cluster 2 (16%) has a UK high at day 11, but then pushes heights away to the east. A period of flat westerlies at day 13, then cooler north-westerlies by day 15 as blocking sets up in the mid-Atlantic towards Greenland.

Cluster 3 (16%) keeps warm weather going with easterlies at day 11, followed by a potential plume setup at day 3 which could be briefly very warm. By day 15, high pressure moves towards Iceland allowing a flat westerly to move in underneath it.

Cluster 4 (16%) has a relatively cool north-easterly at day 11, though unlikely to be quite as cool as the current spell as the direct flow of Arctic air is blocked off to the north around Iceland. The Azores high builds in at day 13 turning things warmer and drier, before a more typical NW/SE split with westerlies sets in at day 15.

Cluster 5 (16%) has a cool north-easterly at day 11 and day 13, but remaining generally dry. By day 15, signs of the high retrogressing further allowing another cool or very cool northerly by day 15.

Cluster 6 (12%) has a strong UK high at day 11 and day 13. Likely still fairly warm given its strength. The high then progresses east by day 15, with a W/E split. A possible plume setup, so could be very warm in eastern areas for a time, but the low is then likely to progress into the UK turning things cooler and more unsettled.

Summary

All in all a pretty clear overall pattern for next week. It does generally look warm through to at least Friday. Possibly very warm for a time depending on the detail of the high orientation.

Beyond that, it's hard to know what to make of the single cluster in the day 8-10 range. This is likely a blend of options that include a full retrogression and ones that keep the easterly going, so I wouldn't make any definite call yet about next weekend.

Looking beyond day 10 seems even more hopeless.   A couple of days ago the retrogression idea had a bit more strength at this timeframe, but it's now a bit less clear. The ensemble is split roughly in three between typical autumnal westerlies, retrogressions, and continued warm conditions with potential plumes. I think we will just have to wait a bit longer for a stronger signal.

The ensemble supports this uncertainty on temperatures towards the end of the month - the mean falls back to average, but no real signal as yet for it to fall below average. And look at that scatter!

 

[Don]

 Daydream Boy Yep, a new thread needed me thinks.

[WYorksWeather]

It wouldn't be the 18z GFS if it didn't give us a bit of a laugh sometimes!

After even an upgrade on next week's warmth compared to the 12z, it then proceeds to go to the opposite extreme with some extremely disturbed weather conditions in FI.

Deep low in the Atlantic at day 11:

A direct strike from a sub-970mb low on the UK at day 13:

A sub-950(!) mb low passing well to our north (fortunately) at day 15:

And here is the situation at day 16, with that deep low just SE of Iceland, and potentially another low approaching from the west.

All a very long way off this, but will be interesting to monitor to see if the GFS is just going off on one this evening. Usually it tends to blow up these lows far too much, but once in a while it gets it right.

[Don]

 WYorksWeather The GFS is a p********r model, even though it gets it right sometimes, it often has to have a laugh!  It's definitely the the rebel of all the models!

Edited September 13 by Don

[WYorksWeather]

 Don I've learned never to totally laugh at it though! Last winter it blew up a 925mb low at long range and sent it to Scandinavia. It ended up being Storm Ingunn, which sent hurricane-force gusts into Norway. That was a ridiculous storm - highest gust was 155mph on the Faroe Islands. Thankfully it didn't affected populated areas too badly.

 

[Don]

 WYorksWeather No, can't rule it out by any means!

Its best merit is that it seems to be good at spotting trends....

Edited September 14 by Don

[Kirkcaldy Weather]

In Depth follow on post

 

[Downburst]

A very settled week approaching. Next Friday ECM and the high still centered around Oslo. Expecting 0.4 mm as the front passes mid afternoon today and no rain for at least a week, maybe 9 days after. Suits me fine I must say.
 

[WYorksWeather]

 Don Probably worth a slightly longer post on this - I see a lot of people post that the GFS seems to pick up trends, so I'll have a look at that idea in a bit more detail below.

I think it's very important not to confuse two different types of question about model accuracy. The question the verification stats answer is: 'given two comparable forecasts at the same lead time, which best represents the state of the atmosphere at the valid time of the forecast?' (i.e. with the least error). So for example, the 7-day forecast verification would compare a forecast from today's (14th) 00z GFS and ECM for 500mb height, for example, to the outcome at midnight on the 21st September, and look at which was closest to reality. On this sort of metric, it is usually the case that the ECM is the best performing model, and the GFS is some way behind, typically between second and fourth.

So, that naturally raises the question of why it so often seems that the GFS spots a trend before the ECM. And the answer is that it simply has more opportunity to do so. The GFS runs four times per day, and the OP runs out to 16 days, whereas the ECM runs twice per day and the OP runs out to 10 days. That means that in any forecast situation, the GFS gets 24 runs (six days' worth) before the ECM gets into range. Frankly, it'd be astonishing if you couldn't pick out a few of those runs that end up being right. Within the day 10 period, there's also more chance for the GFS to 'recover' in our perceptions from an odd-looking run as the next one is only six hours away.

For that reason, when you look back at any particular memorable event, you will nearly always find that there will be a GFS OP run with broadly the correct solution before an ECM OP. However, that does not mean that the GFS is more accurate, simply that it had more opportunities to come up with a correct forecast on at least one of its runs due to frequency and lead time. So, the second question, that we actually ask in hindsight when thinking about an extreme event, is often 'which model was the first to pick up on this?'. And for that, the answer will almost always be the GFS.

In terms of usefulness, I think it should be obvious that the verification stats are a better guide to accuracy than which model picks up on something first. You can easily see this by taking an extreme hypothetical. Supposing we had a model that ran once per hour at very low resolution, and that had a lead time of three weeks. Let's call this Model X. Model X would have verification stats far lower than the ECM or GFS. However, in hindsight, it would almost always seem to pick up on the correct trends before any of the main models, because it would have so much opportunity to do so. But for forecasting, it would be virtually useless, as the accuracy of any particular forecast for a specific point in time would be very low.

 

[Uncertainty]

 WYorksWeather latest run keeps the high going till late next week!

Nevertheless, it looks like support is building for more of a low pressure influence in the extended…

Clusters have Atlantic low pressure making its presence felt late on

Ec46 also keen on squashing the high back into Europe.

Until then, 8 possibly up to 12 or more days of drier weather to enjoy, particularly in NW areas! Only an underestimation of the influence of lp to the south can scupper things, and even that seems less likely at present.

After that? I’ve seen worse OND forecasts from Glosea 5… hopefully we can have a canonical Nina autumn / early winter and a bonus ssw to top things off! I better stop before I get ahead of myself

[raz.org.rain]

 sundog their long range forecasts are interesting, this particular one seems to directly contradict their LRF for winter 24/25 though. Admittedly my respect for that site has gone down a lot since they posted that nonsense AMOC collapse ice age piece.

[Dorsetbred]

18 hours ago, stodge said:

It seems related to the northward shift of the ITCZ which has caused the Sahara rainfall and meant the waves exiting Africa have not developed into tropical depressions/storms despite the conducive environment in terms of SSTs in the tropical central Atlantic and of course the Gulf of Mexico (as we've seen this week with Francine, it only needs a short period of favourable conditions to move a relatively innocuous depression into a Category 2 event).

Why the ITCZ has moved north I don't know -

@stodge by definition the  ITCZ:

The position of the ITCZ varies predictably throughout the year. Although it remains near the equator, the ITCZ moves farther north or south over land than over the oceans because it is drawn toward areas of the warmest surface temperatures. The location of the ITCZ can vary as much as 40° to 45° of latitude north or south of the equator on land. 

[WYorksWeather]

I think we've now reached a pretty good consensus about next week - I've shown the GFS below but most other models are fairly close. Very good agreement on warm high pressure continuing through most or all of next week.

High pressure taking charge by Monday.

By Tuesday, the slight tilt of the high becomes evident. This aids in sourcing the air from more of an ESE direction rather than a straight easterly, and hence it is a fairly warm high.

This remains the case up to Friday as well.

Eventually, by next Sunday, signs of the high pulling to the west slightly. More of a north-easterly, so somewhat cooler, but still dry.

Temperatures continue to show a gradual warmup overall over the next few days up to around Wednesday.

The peak of the warmth is then between Thursday and Saturday, with a cooldown on Sunday for areas further NE exposed to the north-easterly wind.

 

 

[stodge]

Evening all

Signs in far FI of a significant pattern reset with heights moving into the Mediterranean basin from where they have been absent for some considerable time and a return to a more "normal" pattern  with deepening LP around Iceland and a stronger jet - basically what you'd expect as we move in to autumn.

The indications from the PV forecasts I've seen are of a rapid strengthening into October - again, no great surprises. 

Ex-tropical systems can still be players but there's also the likelihood of the long Azores-Europe ridge setting up in response to the restoration of the pattern keeping southern Britain mainly settled and often warm while northern areas are more prone to wind and rain off the Atlantic.