The Doughnut Summer of 2023?

In another notable year, June was the hottest month of Meteorological Summer 2023, the first time that has occurred since 1970. High Summer (July and August) this year have been cooler than in most recent years (and since 2015), but even that would have been above average up to about 1990. September is starting with a heat wave, making 2023 something of a “Doughnut Summer”. There’s even an outside chance of 2023 being the first year in the entire CET (back to 1659) with June and September as the hottest two months!

Well, the data is in:

Fig 1 Mean CET 2023, to date, from the Met Office

As trailed a month ago, June 2023 was indeed the hottest month of at least the meteorological summer. And by a decent margin.

It’s the first time June has been the hottest month or equal hottest in the CET since 1970, which by my reckoning is the longest such gap in the whole record, dating back to 1659.

The Met Office report that June was also the hottest for the UK as a whole:

In a series which goes back to 1884, this is just the seventh time that June has been the warmest or equal-warmest summer month for the UK, with July and August more typically getting the higher temperatures.

Mixed conditions bring warm and wet summer for the UK, Met Office press release, 1/9/23

In the same period (i.e. since 1884) in the CET June was warmest in 1889 (= with July), 1896 (= with July), 1920, 1922, 1940, 1950, 1960, 1966 and the aforementioned 1970. With 2023 that’s 10 years in total. I don’t know which of those (or potentially others) are in the Met Office’s 7 warm June years.

1920 and 1922 were particularly “disappointing” summers, with June the warmest month at 14.4C and 13.8C, respectively! What was going on then? That’s even worth a graph, since I have it handy:

Fig 2 High Summer mean CET temperatures, graph by the author, from Met Office data

So the fairly average June temperatures in 1920 and 1922 (see graph featured in the previous post), were enough to make June the hottest month in those years, in the CET at least.

The other reason for presenting Fig 2, though, is to show that even though High Summer (July and August) 2023 was cool by recent standards, it would have been considered warmer than average any time before around 1990!

The Met Office noted that “[m]eteorological summer 2023 was the eighth warmest on record by mean temperature, thanks largely to June’s record breaking temperatures, in a series which dates back to 1884″ which I have to admit did surprise me a little (given the temperatures in July and August), but just goes to show how much things are warming up.

Obviously (see Fig 2), High Summer 2023 wasn’t the 8th warmest since 1884, but what about the meteorological summer?:

Fig 3 Meteorological Summer mean CET temperatures, graph by the author, from Met Office data

By my reckoning, Meteorological Summer 2023 was “only” 16th= warmest in the CET since 1884. Still, it ranks higher than I would have guessed.

What’s most notable about Fig 3, though, is how rapidly average Meteorological Summer temperatures (indicated by the red and black lines) are increasing. We’ve just seen, by some margin, the warmest 11 and 21 year periods of summers in the entire record, back to 1659.

But what about September?, I hear you ask, having seen in Fig 1 that the mean temperature was 17.5C for the first 3 days. And that was before the heatwave that’s just started:

Fig 4 London temperature forecast, from

The very left of Fig 4 looks a bit odd, since it’s 29C here in West London right now (15:30 on 5th September), so here’s another take:

Fig 5 London temperature forecast from Meteoblue

It’s not expected to be incredibly hot (35C or so was possible in early September even before global warming really set in) but I suspect 30C will be recorded somewhere in the country every day from 4th (Whitechurch 30.2C already reported) through 11th – which, if memory serves will be something of a record – with possible tropical nights (20C+ minimums) to boot.

Could September be even hotter than June? That would make September the hottest month in the year for only the second time (the other was 1890) in the whole CET record, since 1659.

Figs 4 and 5 relate to London, one of the hotter parts of fthe Central England area, but nevertheless, I’d guess that the mean CET for the first third of September is likely to be 5 or 6C, maybe more, above the 1961-90 mean for the month of 13.5C (see Fig 1).

The trouble is, for September to be hotter than June the whole month would need to average 3.5C above the 1961-90 mean. The other 2/3rds of September would still need to be warmer than the historical average for September. And as Figs 4 and 5 illustrate, September is the start of autumn and it cools down.

Furthermore, the CET in September has never averaged as much as 17C:

Fig 6 September mean CET temperatures, graph by the author, from Met Office data

On the other hand, the September CET mean was 16.8C in 2006 and 16.6C in 1729 and there’s been some global warming even since 2006.

Perhaps a more realistic, though still unlikely, record is for September 2023 to be hotter than 16.4C – that is, at least the third hottest in the CET – making 2023 a true Doughnut Summer, with June and September the two hottest months of the year.

Since 2023 would, by my reckoning, be the first Doughnut Summer since at least 1659, I’ll be keeping a close eye on how September pans out. I won’t be putting any money on it, though!