4 Comments
User's avatar
Dave Woolcock's avatar

Are your LOESS curves correlated at all? First thing I noticed was that all showed a rise in the final 15-odd years. Would it mean anything if they were? (Thinking about your comments on methodology and paper cups)

Expand full comment
John Dee's avatar

Not only are the LOESS curves correlated but the raw data series also. In fact, the raw data yields stronger correlation coefficients than anything smoothed, with the highest value of r = 0.758 (p<0.001) being attained by the South West and South East England series. This makes total sense since dirty great weather fronts can swamp our tiny island, generally moving West to East. The weakest correlation is r = 0.335 (p<0.001) between Central England and the North West, but there’s still a positive correlation. This is exactly what I’d expect.

What we are riding pretty much everywhere is a recent bump from 2005 onward that I’m guessing will flatten out over the next decade or so. There’s no evidence of underlying long-term trends but what I have gone and forgotten is to include the Wald-Wolfowitz runs test results that show all 8 series to be a random walk. I’ll update the article with these when I can grab a moment (currently cooking a roast chicken dinner!).

Expand full comment
Dave Woolcock's avatar

not surprised by the NW low correlation score - that's where I live. Somewhere I have a satellite pic of the whole of the UK covered in snow.... apart from one little area in the NW of England.

Expand full comment
John Dee's avatar

There you go! Always nice to put flesh on the bones.

Expand full comment