Mark Smithy

Although the models are consistent about a cool down, the 18z GFS operational was a cold outlier in FI: What does seem to be the case is that we're in a transition phase. Signs of the hurricane season finally awakening too: https://www.accuweather.com/en/hurricane/is-the-tropical-atlantic-on-the-verge-of-turning-active/1236418

Goodness me: If this was winter this place would be in meltdown: And why do charts like that never come off in winter itself?!

Well to be fair the 0z GFS operational is also an outlier for lack of precipitation! I think there's a slight tendency at the moment, if I may say so, to do the same thing that we cold lovers have to beware of in winter: i.e. reading into model outputs what we want to see and selectively filtering out the outputs we don't like the look of. September can indeed be a benign month synoptically but northerlies can occur at any time of year.

Yes BBC / Meteo Group just now said that the westerly extent of the rain is still open to doubt and that "it might clip the East Midlands", which suggests they are siding more with ECM / UKMO: The track of these kinds of events is as you say often a case of nowcasting but I don't trust the GFS on precipitation at the moment. Imho it has been really inaccurate lately.

Maybe something which looks like a polar low synoptically but isn't entirely that without colder upper air temps. The profile of a low pushing down on a north to south trajectory whilst deepening. However, I did say I thought this was quite likely to disappear from the outputs because in winter they invariably downgrade! The models look to be struggling a bit at the moment.

Well to be strictly accurate you said it was very unlikely. I can't really see why it should be so unlikely? If the high retrogrades into mid-Atlantic as some model outputs seem to suggest then that does indeed open the door to a northerly. Swings and roundabouts really. I'd not be surprised by the odd cooler incursion in September / October alternating with warmer spells.

So the Met Office have just confirmed that 26.8C recorded at Shirburn model farm on 19th July 2022 is the highest minimum temperature ever recorded in the United Kingdom. Not entirely sure why it took a month to verify but anyway.

Did you mean to write a cool northerly is very unlikely at this stage? Otherwise there's not the blindest reason why a northerly can't occur at any time of year. There have been some notable examples even in summer time.

Certainly a cool set up here from the 0z GFS in following the ECM's lead. The 18z also had it in potent form: However, it's a long way off and it's the kind of brief set up which almost always downgrades in winter. So those looking for a cool flow in September, I wouldn't necessarily hold your breath.

Yes just at the moment that the 0z GFS decides to follow the ECM with the low pressure troughing (almost a polar low!), the UKMO throws in that set up. Could be glorious (if you like late summer warmth). Looking to the bank holiday weekend, the GFS has bowed to what looked like the inevitable given its own ensembles and it has now gone with the slightly cooler flow off the north sea. Pleasant temps in the south but nothing too hot. And the potential still for some decent rain tomorrow across the south east.

Amazing differences still in the model mix at such short notice for this weekend. The 06z GFS was a warm outlier for the relevant time and there's a lot of scatter at short range here:

Not a lot of support in the GFS ensembles for a hot bank holiday weekend. The majority of GFS members are now following the ECM solution. Pleasant enough though: Interestingly though, there is significant support for the rain in the south-east on Thursday:

Actually think it has been far more accurate over the past fortnight than the GFS which is the one that has performed less well. In terms of maxima the GFS got the last hot spell wrong every day by c. 2 to 3C. On precipitation it has been terrible: failing to pick up on the admittedly subtle changes going on, and it has switched back and forth on positionings of high pressure. The latest fiasco is that up until two days ago it had us sitting under high pressure over the bank holiday with temps of 30C. Now it has a slack flow with charts on the 18z and 0z that don't really match any other forecasting model: all of which have a gently cooling east to northerly flow across the southeast over the weekend. To be reasonable to ALL the model outputs, there is no real impetus from the Atlantic so in such slack flows the positioning of high pressure cells by a couple of hundred miles can make all the difference.

Ooops. That's quite the error on my part And no amount of digging is going to get me out of it. However, I still think that September is a late summer month, or a month which has many summer features about it and on that level the equinox of September 21st feels about right. The more the warming continues, the more we may have to consider altering these kinds of traditional seasons.

There's some pretty heavy rain coming into the south-west at the moment. We will recall that the GFS has had virtually no rain showing, which I don't really understand tbh, unless this fizzles out which I guess it might?

No I'm afraid I fundamentally disagree with this and always have (so it's not just you). I think you have let the personal cat out the bag when you write of earlier sunsets. I get that, but with respect that's a psychological, not a meteorological, observation. September is a late summer month. It's one of the occasions when equinoxal assessment wins out (i.e. summer until Sept 21st). September is far warmer, for example, than May which is classed as meteorological summer and it's only just behind June! The highest maxima for September is 35.6C and 34.4C was reached as late as 13th September in 2016. 34C has been breached on several occasions in September. The highest May temperature by comparison is 32.8C CET means: May 11.1C June 14.1C July 16.0C August 15.8C September 13.6C Objectively September is a much warmer month than May as borne out by the statistics. p.s. try working on a mind reset about this? Look for summer signs into October and again in those early glimpses in February and March. It can work wonders for the mental health.

It would be quite extraordinary were summer 'done with' on August 22nd. September usually has many hot days, with 30C recorded on the majority of days in September at one time or other: The Tornado and Storm Research Organisation (TORRO) WWW.TORRO.ORG.UK TORRO is a privately-supported research body specialising in severe convective weather in Britain and Ireland We have even come within 0.1C of 30C on October 1st and with climate change I wouldn't bet against that happening at some point in the not too distant future.

Models are continuing to struggle under the slack flow at the moment. None of them seem quite sure where the weak high pressure is heading, including two different scenarios between the GFS 18z and the 0z: To be noted that the 0z GFS is above the mean on temperature: And WAY out of kilter on precipitation: But the GFS 0z does have some support from the UKMO which shows a weak nose of high pressure extending out of the Azores for the bank holiday: Of course, we don't know where the UKMO would head next but the ECM has the weak high pressure retrograding, to leave a cool flow across the UK, including for the bank holiday: Finally, the FAX charts, which are imho always the go-to for short range forecasting, have two weak weather fronts crossing the country on Friday: What to make of all this? The bank holiday looks okay-ish in the south-east but not a heatwave and it may be quite cloudy at times even down south. The 0z GFS was a lot warmer for the south-east than the 18z, which favoured the 0z ECM solution i.e. fairly cool winds off the north sea: Given the fairly weak support for the warmer 0z operational run, the ECM is the likelier of the two but that doesn't mean the latest GFS won't be right. As I mentioned yesterday, nothing amazing on heat, cold, or precipitation.

Yes and quite a disparity again this evening between GFS and ECM. Generally at the moment I see nothing much in the models that on the one hand warrants the suggestion of early autumn nor on the other the suggestion of a heatwave returning. Generally the models, despite their differences, are throwing up fairly typical late summer mixed conditions. Some sunshine. Some warm, humid, days. Some precipitation especially away from the south east. Nothing extreme.

The GFS is dry on its operational again and so are most of the ensembles. Looking ahead the operational remains very dry but not some of the other members: The ECM and BBC / Meteo certainly have a fair amount of ppn showing for tomorrow:

Meanwhile, the ECM looks good for those who would like to see southern England return to being a green and pleasant land Some much-needed precipitation through the run, with the Azores high being kept at bay to our south-west, allowing an Atlantic flow across the UK:

FAX charts also show a very unstable situation going into next week. This set up combined with residual heat will produce a lot of precipitation for some. I always favour the FAX charts, especially over the North American model when it comes to pinpointing precipitation. Nick Finnis' Netweather update this morning also looks to have this spot on: Peak of the heatwave today but the end is in sight with some welcome rain WWW.NETWEATHER.TV Heatwave will reach its peak today, with temperatures perhaps reaching 37C in southern England. Hot again Sunday before heat eases and threat of thundery downpours increases as we head through...

The 0z ECM is a pretty decently wet run throughout. Looking good for the parched south. Some sample screenprints: Given that the Euro models are in agreement at the moment, as are most of the GFS ensemble members, it's the GFS operational which continues to be operating in an inaccurate world of its own. It has stuffed up on the maxima throughout this spell, so its latest outlying modelling should be taken with a health warning. Unless of course its own members and the Euro models come on board, in which case different story. For now? Be sceptics.

12z GFS operational is a drought outlier: Something that's really odd about the operational at the moment is that it doesn't even start in the same place as the rest of its members. It actually begins 2C warmer than the rest. No wonder it keeps showing temps that are 2C too high if it can't even get out of the blocks right.

In assessing the performance of the GFS over the next few days it should be noted that yesterday it over-baked temperatures for many areas by 2C and it has done so again today. It should be noted that this is true of the GFS operational runs. As I have repeatedly demonstrated, for some reason the GFS operationals have been running about 2C higher than the vast majority of its own members, even at only 6 hrs range. I don't know what's wrong with the GFS but it has a problem.