How Not to Report a Weather Record: 21st June 2017
Well, well, well. Less than a year on from an exceptionally hot mid-September day (at least exceptionally hot for the UK, if not, perhaps, for Kuwait), and it’s only gone and happened again.
Yeap, the presumably less poisonous than mercury red liquid in my re-purposed fridge thermometer has only gone and reached 34.5C this week, on what was widely reported as “the hottest June day for 41 years”, that is, since the summer of 1976. And curiously I was close to the epicentre of the heatwave back in ’76, in FA Cup-winning Southampton, then the hottest place in the country, just as where I am now, a few miles from Heathrow, has been this time.
And once again the record has been somewhat understated. I explained in my post on the topic last September that the true significance of the 13th September 2016 was that it was the hottest day that had been recorded in the UK so late in the year.
You’ve guessed it. The 34.5C recorded at Heathrow this summer solstice was the hottest daily maximum so early in the year. Back in 1976 the temperatures over 35C (peaking at 35.6C in Southampton on 28th) were later in the month. In other words, 21st June 2017 saw a new “date record”.
Admittedly, it was not a particularly notable date record, since 34.4C was recorded at Waddington as early as 3rd June during the glorious post-war summer of the baby-boom year of 1947. And 35.4C at North Heath on 26th June 1976 also seems somewhat more significant than nearly a whole degree less on 21st June. Furthermore, unlike in 1947, 1976, and, for that matter, 1893, only one “daily record” (the hottest maximum for a particular date) was set in the 2017 June heatwave.
Nevertheless, 21st June 2017 set a new date record for 5 days (21st to 25th June, inclusive) and that is of statistical significance. The point is that without global warming you would expect there to be approximately the same number of date records each year, or, more practically, decade. The same is true of daily records, of course – providing a recognised statistical demonstration of global warming – but my innovation of date records provides for a more efficient analysis, since it takes account of the significance of daily records compared to those on neighbouring dates. It makes use of more information in the data.
Supporting the “hypothesis” of warming temperatures, the 5 day date record set on 21st June 2017 exceeds what you would expect in an average year, given that daily temperature records go back over 150 years. On average you’d expect less than 3 days of date records in any given year. But we can’t read too much into one weather event, so how does it look for recent decades?
Last September, I provided a list of UK date records from the hottest day, 10th August, when 38.1C was recorded in Gravesend in 2003 through to October 18th, promising to do some more work next time there was a heatwave. So, keeping my word, we have the following date records:
34.4C – 3rd June 1947 – 18(!) days
34.5C – 21st June 2017 – 5 days
35.4C – 26th June 1976 – 1 day
35.5C – 27th June 1976 – 1 day
35.6C – 28th June 1976 – 3 days
36.7C – 1st July 2015 – 33(!!) days
37.1C – 3rd August 1990 – 7 days (through 9th August)
Obviously, weighting for how exceptionally hot they were, the 2010s have had way, way over their share of exceptionally hot days for the time of year during the summer months. I’m timed out for today, but I will definitely have to get round to an analysis of the whole year! Watch this space.
Just discovered your blog, your analysis of date records is fascinating and different to what anyone else has done- thank you and keep it up!