Referring back to my post below from the Sept CET thread it is interesting to note that of the three years that have all the following characteristics - CET for 21-30 Sept between 10.5C and 11.5C, very low solar activity and summer CET of 16C+, the CET for the start of October 2018 is also similar to these earlier three years.
The CET for 1-13 Oct is 13.09C by my calculations. Of the years in the table below there are not many warm Octobers. But the three matches I refer to above - 1878, 1807 and 1822, all three had warm starts to October.
1878 saw a CET of 12.63C to the 13th, 1807 had 12.99C and 1822 12.31C. None quite as warm as this year but still close enough.
All these years had very cold winters: 0.7C in 1878, 2.4C in 1807 and 1.5C in 1822. November was also very cold in two years but quite mild in 1822.
Worth noting, if I recall correctly, that 1878 is coming up a lot in Gavin P's winter analogues this year. It does seem to be a very good match indeed. That was one of the coldest winters on record. This is all speculative of course but interesting nonetheless.
Just looking at when the cold weather set in during those three years, in 1878 the CET went sub 5C on 29 October and apart from a handful of days in November and again in February it stayed below 5C until early March. There are signs of much colder weather potentially kicking in this year in the final few days of October.
In 1807 the cold did not really kick in until 8 November. In 1822 it was on 29 November.
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.
IMAGE. Members enable at bottom of page
TABLE1
IMAGE. Members enable at bottom of page
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.