A Modest Cloud Cover Study (part 7)

Today I boil the monthly values for the CRU TS4.08 and corrected ICOADS v3 datasets down into annual means and have a look-see at these

As the subtitle states I am now going to get the big stock pot out and boil all that monthly data down until we are left with two tender time series of annual mean values. Tenderising data using the arithmetical mean of something dominates science, particularly popular science destined for public consumption yet, as I have shown in previous articles, this summary measure can easily misinform and mislead. It’s the perfect political weapon, and I urge readers to always ask what the mean of anything actually means.

If I measure the height of a bunch of five-year-old kids, then measure the height of a bunch of fifty-year-old adults, then I can derive a single mean value for what I’ve measured. If I then go and manufacture 100% cotton dinosaur motif aprons using that mean value then I’m going to produce a whole bunch of groovy kitchen gear that will be utterly unsuitable for purpose (apart from fitting the odd child and adult that just so happens to be close to the mean height of both groups). Of all the applied fields I’ve worked in as a statistician these last 40 years I must say that climate science takes the biscuit when it comes to ill-fitting aprons!

So, then, let us sip our cuppas and have a squizz at the mean annual values for cloud cover for the modified ICOADS v3 (2°) and as-it-comes CRU TS4.08 (0.5°) datasets for the region of interest. When I say ‘modified ICOADS’ I mean adjusted for the wartime wacko dip that runs from 1945 – 1956 (see part 4).

Here we go…