Studying the variability of UK temperatures

[Roger J Smith]

Using an excel file containing all the daily data from 1772 to present (CET series), I derived the co-efficient of variance for each year to study how variability in the temperature record has changed or otherwise behaved over time.

 

The numbers in the table have no absolute significance, but can be compared to each other to assess changes in the variability of temperature. The higher the numbers, the more variable the temperatures in the given year.

 

This table shows all the annual averages from 1772 to 2013. The top ten are in bold red and the bottom ten (least variable) are in blue and underlined.

 

DECADE ___ Yr 0 ___ 1 ___ 2 ___ 3 ___ 4 ___ 5 ___ 6 ___ 7 ___ 8 ___ 9 _____ median

 

1770s _______ ---- _ ----  _ 791 _ 835 _ 623 _ 707 _ 816 _ 726 _ 766 _ 775 ____ 771

1780s _______ 859 _ 641 _ 763 _ 820 _ 878 _ 797 _ 793 _ 544 _ 774 _ 635 ____ 784

1790s _______ 493 _ 624 _ 757 _ 560 _ 606 _1153_886 _ 561 _ 789 _ 594 ____ 615

 

1800s _______ 617 _ 711 _ 664 _ 881 _ 832 _ 592 _ 545 _ 939_ 905 _ 680 ____ 696

1810s _______ 684 _ 595 _ 701 _ 842 _ 890 _ 787 _ 670 _ 845 _ 765 _ 816 ____ 776

1820s _______ 887 _ 780 _ 639 _ 728 _ 682 _ 677 _ 906 _ 805 _ 730 _ 629 ____ 729

1830s _______ 970_ 693 _ 461 _ 770 _ 591 _ 702 _ 666 _ 809 _ 855 _ 700 ____ 701

1840s _______ 791 _ 881 _ 690 _ 901 _ 689 _ 769 _ 881 _ 763 _ 762 _ 711 ____ 766

 

1850s _______ 717 _ 599 _ 711 _ 670 _ 642 _ 861 _ 757 _ 693 _ 753 _ 954 ___ 714

1860s _______ 509 _ 687 _ 788 _ 597 _ 751 _ 793 _ 697 _1066_739 _ 896 ____ 745

1870s _______ 954_ 764 _ 789 _ 746 _ 879 _ 681 _ 828 _ 690 _ 920_ 640 ____ 777

1880s _______ 738 _1035_549 _ 604 _ 694 _ 551 _ 733 _ 681 _ 883 _ 616 ____ 688

1890s _______ 821 _ 674 _ 888 _ 757 _ 703 _1043_673 _ 649 _ 770 _ 845 ____ 764

 

1900s _______ 703 _ 722 _ 810 _ 760 _ 601 _ 553 _ 702 _ 649 _ 808 _ 702 ____ 703

1910s _______ 605 _ 678 _ 902 _ 519 _ 587 _ 611 _ 782 _ 890 _ 740 _ 799 ____ 709

1920s _______ 705 _ 793 _ 736 _ 772 _ 581 _ 781 _ 842 _ 686 _ 736 _ 879 ____ 736

1930s _______ 610 _ 614 _ 665 _ 712 _ 594 _ 714 _ 688 _ 583 _ 659 _ 645 ____ 652

1940s _______ 761 _ 771 _ 786 _ 502 _ 620 _ 821 _ 716 _1293_868 _ 595 ____ 766

 

1950s _______ 685 _ 472 _ 809 _ 544 _ 741 _ 877 _ 760 _ 665 _ 603 _ 629 ____ 675

1960s _______ 555 _ 669 _ 642 _ 842 _ 647 _ 608 _ 619 _ 533 _ 666 _ 737 ____ 657

1970s _______ 750 _ 607 _ 493 _ 530 _ 608 _ 667 _ 830 _ 614 _ 801 _ 763 ____ 641

1980s _______ 709 _ 842 _ 697 _ 749 _ 576 _ 863 _ 786 _ 820 _ 566 _ 651 ____ 729

1990s _______ 718 _ 793 _ 627 _ 636 _ 670 _ 811 _ 671 _ 698 _ 630 _ 555 ____ 671

 

2000s _______ 567 _ 655 _ 524 _ 637 _ 561 _ 740 _ 760 _ 647 _ 644 _ 597 ____ 641

2010s _______ 905 _ 738 _ 768 _ 759 _ (334)

 

 

Just a few notes on these variability figures. The historical trend has been somewhat downward from a peak in the late 19th century to minima around the 1970s and the decade of the 2000s. The pace had begun to pick up again in 2010 but has fallen off sharply so far in 2014. Given that the winter months tend to have higher averages than the summer months in this statistic, the January to March average of 334 is considered exceptionally low.

 

 

If you scan the data for red and blue figures to find the top and bottom ten cases, you'll find that there has not been a top ten year since 1947 which was first overall. 2010 fell just outside the top ten, and many of the years in this set have had a cold winter, for example, 1795 and 1895 as well as 1830 and 1881. Each decade from the 1850s to the 1890s had at least one of the top ten variability years.

 

The least variable years cluster more recently, in particular from 1964 to 1975, a 12-year interval which failed to produce a notably above-average year. 1832, the least variable year, seems somewhat out of place in the data, as there was no other bottom ten case from 1791 to 1859. The frequency of low-variability years picks up in the 20th century and 1951 is the least variable year in recent times, with 1972-73 producing consecutive bottom ten readings, and 2002 the most recent extreme case.

 

In general terms, the evidence here is not very strong for the oft-repeated mantra that with climate change, a more variable climate is developing, that only seems to be true after 2009 and otherwise the AGW period (from about 1982 on) has mixed results that are not generally higher than the long-term average of variability. The most variable decades would appear to be the 1860s and 1940s, taken together with the 1770s, this indicates a roughly 80 year cycle of higher variability.

 

 

I may post more after reviewing the data and any comments or questions that arise. Note that the variance statistic will highlight years with a wide range of temperatures more than a year with consistently anomalous temperatures in one direction (such as a consistently warm year). The statistic is looking for a range of anomaly values rather than large anomalies all of the same sign.

 

 

It would be interesting to compare North American daily data, my subjective impression has always been that the 1970s provided a peak in temperature variability in the eastern part of North America, especially 1976-77. A less variable period was likely found from 1989 to 1992.