Hi Gusty, I wonder if you could elaborate on this? 

You see, I'm puzzled by your statement, as when performing correlation analysis, the August temperature trend (how much it declines and how fast) shows no significant relationship with how September pans out. Given that the consensus of model guidance and historical precedent (regarding tropical cycle timing and impact on weather patterns; the July-September tropical progression this year, in the face of ample sub-surface oceanic heat and following a dual-year Nina-like period, is a first in at least the modern records) was strongly in favour of month between 1 and 2*C above the long-term average, I am very intrigued as to your line of thought here?

Cheers

As for going forward from here, well that is some truly fascinating analysis by GW in recent days, and while purely statistical it does make a good argument for cooler than usual conditions to predominate for the next half-year, give or take a month perhaps. 

The tropical cycle behaving so strangely is lowering confidence in all forms of historical data-based analysis at the moment, but as this analysis takes into account the odd September result, it may well be the best guidance we have, at least for the time being. Well done I say!

There are strong anticyclonic signals developing in the model output for October and, with the shortening days, this does support a cooler than average CET overall, even if some of the days turn out warm.

Hi SC. Nothing statistical.

For me it 'felt' very much as if the onset to autumn started around the 9th August. (Very early locally). Call it a good old fashion hunch. 

The summer never recovered thereafter and the trend has continued as we now move deeper into autumn. 

Ah well okay, you know what, hunches work as well as anything in a game of probability so fair enough I suppose. After all, that's all I was using back in 2010... alas, there's been plenty of evidence since then that I just got very lucky! 

I've not been alive long enough for an early start to an autumnal feel to mean much to my 'gut' so to speak; it's been so rare down here with even the more disappointing Augusts still tending to be mild enough by night to be above average. This year being no exception.

Today's GFS runs have trended cloudier for later this week with higher minimums - enough so that I get a final estimate from the 12z of 13.9*C.

Trouble is, I'm not very confident in GFS' predictions for tonight and tomorrow night; the numbers seem 1-3*C too high based on past experience of the setup we currently have.

Maybe a chance, though, that Wed and Thu can reach a bit above what the model is going for, to offset any lower temps the next two nights. 

If only that new push east from the ridge would happen a bit faster during Fri - that'd send most of the cool polar maritime air into the North Sea instead of drawing it across most of the UK.

Met Office Hadley      14.4c.      Anomaly     0.5c. Provisional to 24th.

Metcheck                   14.15c     Anomaly     0.43c

Netweather                14.24c     Anomaly     1.15c

Mansfield                   14.0c       Anomaly     0.22c

Peasedown st John      14.49c.   Anomaly     -0.48c

All 10 down today

Mean of  my 10 stations 14.50, 

Met Office  Hadley       No Update

Metcheck                    14.09c    Anomaly     0.37c

Netweather                 14.66c    Anomaly     0.97c

Hexam                        13.1c     Anomaly     0.06c.

Mansfield                     14.9c     Anomaly     1.12c

Peasedown st John 14.38c  Anomaly   0.64c

Treviskey Redruth   14.5c   Anomaly   0.24c

3 Up 7 Down

Average of my 10 stations  14.42c.    

I think I perform better when I forget to enter!

By my calculations the CET currently stands at 14.05C. 

Today's CET was 1C above average and tomorrow should be the same. After that it turns cool for the final 3 days of the month.

My current estimate for the final September CET is back down to 13.70C.

Looks like I’ll be way too high then!    

I wonder if I can do the same next month.   GW, I’m off on holiday again, so I’ve PM’d my October guess.  

Met Office Hadley      14.2c.     Anomaly     0.4c. Provisional to 26th.

Metcheck                   14.17c     Anomaly     0.45c

Netweather                14.65c     Anomaly     0.96c

Mansfield Weather      15.2c       Anomaly     1.42c

Peasedown St John     14.31c     Anomaly     0.66c

Treviskey Redruth       14.6c      Anomaly     -0.14c.

1 Down TWO Steady    7UP

Mean of my 10 watched stations  14.56c

Thanks GW. 

Looks like temperatures have cooled down to normal after that protracted hot spell from April to July.

Wonder where we go from here for Q4?

Met Office Hadley        14.3c.     Anomaly     0.5c.  Provisional to 27th

Metcheck                     14.18c    Anomaly     0.46c

Netweather                  14.68c    Anomaly     0.99c

Alton                           14.1c      Anomaly     -1.16c

Cheadle Hulme             13.9c     Anomaly      -0.14c

Hexam                         12.9c     Anomaly      -0.26c

Mansfield                      15.0c     Anomaly      1.22c

Peasedown St John         14.32c    Anomaly    0.65c

1 Steady  TWO up  7 Down  5 Below their 5 year average.

Mean of my watched 10 stations   14.43c.    

Met Office Hadley        14.2c       Anomaly      0.5c  Provisional to 28th.

Metcheck                     13.99c     Anomaly      0.27c

Netweather                  14.53c     Anomaly      0.88c

Alton                            13.9c      Anomaly      -1.36c

Hexam                          12.8c      Anomaly      -0.36c

Mansfield                       14.9c      Anomaly      1.12c

Peasedown St John         14.20c    Anomaly      -0.77c  

Treviskey Redruth           14.3c.     Anomaly      -0.44c.

10 Down Today 

9 Below the 5 year average

Mean of my 10 watched stations   14.3c 

I did some further work on looking at the impact of cool late Septembers on the following winters. I have now extended the analysis back to 1800 and also included years where the CET for 21-30 Sept is 11.45C and below. The figure for this year is likely to be around 10.95C so the really cold late Septembers are not such a good match.

TABLE1

TABLE2

What this undated analysis shows is that although a cold end to September usually results in a very cold October, a somewhat less cold end to September actually results in October being less cold, although still below average. Where the late September CET is between 10.45C and 11.45C the following October comes in at 9.6C on average. Whereas if the late September CET is below 10.5C the following October comes in at 8.9C on average.

The extended analysis brings in a lot more years because there are many years in the first half of the 1800s that match the criteria. If we just look at the years with low solar activity we see that it does not make much difference compared to all the years put together. The exception is December which is much less cold, in fact slightly above average. January is also closer to average.

But if we consider years with very low solar activity, i.e. close to solar minimum we see a very different picture. December and January become very cold indeed, as does March. October is less cold and other months are largely unchanged. The data set for solar minimum years is actually reasonably large with 8 matching years having cool ends to September. They are 1808, 1824, 1809, 1822, 1807, 1888, 1878, 1811. The downside is all these years are a very long time ago and mostly in a small window between 1807 and 1824. So may not be a good comparison for today.

This updated list also has a much larger number of years with very warm summers. If we look at summers with a CET of 15.5C or more there is little difference to the overall average temperature by month. But if we look at years with 16C or more (there are 9 matches) we see that, as for low solar activity, December and January become very cold indeed and all months, apart from October are colder with an anomaly of at least 1.3C.

Combining years with a 16C+ summer CET and late September CET between 10.45C and 11.45C we have 6 matches. This shows an interesting trend. October is now slightly above average compared to the overall mean for all years which showed a negative anomaly of more than 1C. But November through to February all show very negative anomalies of 1.5C or more, with January at -3C. Finally if we add the low solar criteria (low rather than very low) we have just 3 matches. This includes one modern year, 1990, as well as two from long ago, 1878 and 1807. The data here are very stark. October is again just above average but all Nov, Dec, Jan and Mar all have negative anomalies of 2C or more, with December and January above 3C. The overall winter CET is just 1.6C. Now of course there are only 3 data points here and so this should be taken with a bit of a pinch of salt. 1990 is perhaps the year to focus on. This actually had a very mild Oct and Mar with an average Nov. The winter was cold though, especially February with the overall CET 1.5C below average for the winter. 1807 and 1878 were both incredibly cold from November right through to March.

So some interesting data there I think. Based on those years that are the closest match to 2018, the data would suggest that actually October may be a fairly average month. But after that things could turn quite cold and potentially stay that way for some time. All just for fun of course but interesting nonetheless.

Latest CET estimate for September is 13.68C

Thanks Simon for such excellent analysis and data there.

Met Office Hadley          14.0c.      Anomaly      0.4c  Provisional to  29th.

Metcheck                       13.87c.    Anomaly      0.15c

Netweather                    14.4c       Anomaly      0.71c

Alton                             13.8c       Anomaly      -1.46c    All my 10 stations using a 5 year average of their own.

Canvey Island                15.3c       Anomaly      -0.5c

Cheadle Hulme               13.7c       Anomaly      -0.34c

Clevedon Weather           15.3c      Anomaly       -0.82c

Darwen                          11.8c      Anomaly       -1.36c

Hexam                           12.8c      Anomaly       -0.36c

Mount Sorrel                   14.0c      Anomaly       -2.76c

Mansfield Weather           14.8c      Anomaly       1.02c

Peasedown St John          14.09c    Anomaly        -0.88c

Treviskey Redruth             14.3c     Anomaly       -0.44c. 

Mean of my watched 10 stations   14.03c.

Fascinating analysis GW.

Looks like Sep will end fairly close to the 1981-2010 average; my guess was about a degree too high this month.

Good range to compare by Art; I look forward to your final list tomorrow.