How Not to Report a Weather Record: 13th September 2016

Last Sunday, the Guardian website suggested Tuesday 13th September would be jolly warm:

“If the mercury rises above 31.6C, the temperature was [sic] reached at Gatwick on 2 September 1961, it will be the hottest September day for 55 years.”

“No, no, no!!”, I was obliged to point out, adding, by way of explanation that:

“If the temperature rises above 31.6C it will be the hottest September day for more than 55 years, since 1961 was 55 years ago.

For it to be the hottest September day for 55 years it will only have to be hotter on Tuesday than the hottest September day since 1961.”

Good grief.

After that I was hardly surprised – since your average journo seems not even to be an average Joe, but, to be blunt, an innumerate plagiarist – to read in the Evening Standard on the 13th itself:

“If the heat rises above 31.6C, which was reached at Gatwick on September 2, 1961, then it will be the hottest [September] day for 55 years.”

See what they’ve done there?  With a bit of help from Mr Google, of course.

In the event, it reached 34.4C on 13th, making it the hottest September day for 105 years.

Much was also made of the fact that we had 3 days in a row last week when the temperature broke 30C for the first time in September in 87 years.

But the significance of the 34.4C last Tuesday was understated.

The important record was that the temperature last Tuesday was the highest ever recorded so late in the year, since the only higher temperatures – 34.6C on 8th September 1911 (the year of the “Perfect Summer”, with the word “Perfect” used as in “Perfect Storm”) and 35.0C on 1st rising to 35.6C on 2nd during the Great Heatwave of 1906 – all occurred earlier in the month.  By the way, in 1906 it also reached 34.2C on 3rd September.  That’s 3 days in a row over 34C.  Take that 2016.  They recorded 34.9C on 31st August 1906 to boot, as they might well have put it back then.

No, what’s really significant this year is that we now know it’s possible for the temperature to reach 34.4C as late as 13th September which we didn’t know before.

I’m going to call this a “date record”, for want of a better term.  Any date record suggests either a once in 140 years freak event (since daily temperature records go back that far, according to my trusty copy of The Wrong Kind of Snow) or that it’s getting warmer.

One way to demonstrate global warming statistically is to analyse the distribution of record daily temperatures, i.e. the hottest 1st Jan, 2nd Jan and so on.  Now, if the climate has remained stable, you’d expect these daily records to be evenly distributed over time, a similar number each decade, for example, since 1875 when the records were first properly kept.  But if the climate is warming you’d expect more such records in recent decades.  I haven’t carried out the exercise, but I’d be surprised if we haven’t had more daily records per decade since 1990, say, than in the previous 115 years.

It occurs to me that another, perhaps quicker, way to carry out a similar exercise would be to look at the date records.  You’d score these based on how many days they apply for.  For example, the 34.4C on 13th September 2016 is also higher than the record daily temperatures for 12th, 11th, 10th and 9th September, back to that 34.6C on 8th September 1911.  So 13th September 2016 “scores” 5 days.

Here’s a list of date records starting with the highest temperature ever recorded in the UK:

38.1C – 10th August 2003 – counts for 1 day, since, in the absence of any evidence to the contrary, we have to assume 10th August is the day when it “should” be hottest

36.1C – 19th August 1932 – 9 days

35.6C – 2nd September 1906 – 14 days

34.6C – 8th September 1911 – 6 days

34.4C – 13th September 2016 – 5 days

31.9C – 17th September 1898 – 4 days

31.7C – 19th September 1926 – 2 days

30.6C – 25th September 1895 – 6 days

30.6C – 27th September 1895 – 2 days

29.9C – 1st October 2011 – 4 days

29.3C – 2nd October 2011 – 1 day

28.9C – 5th October 1921 – 3 days

28.9C – 6th October 1921 – 1 day

27.8C – 9th October 1921 – 3 days

25.9C – 18th October 1997 – 9 days

And you could also compile a list of date records going back from 10th August, i.e. the earliest in the year given temperatures have been reached.

The list above covers a late summer/early autumn sample of just 70 days, but you can see already that the current decade accounts for 10 of those days, that is, around 14%, during 5% of the years.  The 2000s equal and the 1990s exceed expectations in this very unscientific exercise.

Obviously I need to analyse the whole year to draw firmer conclusions.  Maybe I’ll do that and report back, next time a heatwave grabs my attention.

It’s also interesting to note that the “freakiest” day in the series was 2nd September 1906, with a daily record temperature hotter than for any of the previous 13 days.  2nd freakiest was 19th August 1932 – suggesting (together with 2nd September 1906) that perhaps the real story is an absence of late August heatwaves in the global warming era – joint with 18th October 1997, a hot day perhaps made more extreme by climate change.

Am I just playing with numbers?  Or is there a serious reason for this exercise?

You bet there is.

I strongly suspect that there’s now the potential for a sustained UK summer heatwave with many days in the high 30Cs.  A “Perfect Summer” turbocharged by global warming could be seriously problematic.  I breathe a sigh of relief every year we dodge the bullet.