By Paul Homewood
Whenever the Met Office publish their UK temperature charts, they never show any margins of error.
This in itself is poor statistical practice.
And as we now know, most of the Met Office’s temperature recording network has very high levels of uncertainty because of poor siting.
Class 3 measurements are only accurate to within a degree, according to the WMO. Class 4 and 5s, which account for 77.9% of weather stations, have uncertainty up to 2C and 5C respectively.
If we average this together, it means that poor siting could be artificially overstating the Met Office’s UK temperature averages by 2.5C.
Usually in other fields, negative and positive errors would tend to cancel out, due to their random nature.
With temperature recording, however, poor siting nearly always adds to underlying temperatures. If the Met Office had actually published these error margins, their annual temperatures would have looked like this:
I am not saying that the error margin is necessarily as big as 2.5C, only that in theory it might be. But with the predominance of poorly sited stations, it is abundantly clear that the Met Office cannot scientifically claim to know the current average temperature of the UK to a hundredth of a degree.
For the sake of this exercise, I have only shown the error margins for the years since 2010, as we can reasonably assume the existing mix of stations applies throughout that period.
Going back in time, of course, we have no idea of the mix of stations or how badly sited they were. Nor, more importantly, do the Met Office.
That of course means that they have no means of knowing whether they are comparing like with like, when they publish temperature trends going back to 1884.
And they therefore cannot say with any degree of scientific certainty that the last two years were the warmest on record, nor quantify how much, if any, the climate has warmed since 1884.