Lumos' Little Climate Joke

Our good buddy Lubosh Motl has posted some figures on warming trends in Central England from 1659 to the present. When he plots the warming trend (essentially the first time derivative of the temperature) he finds quite amazingly that the past and present look rather alike to the casual eye. Anyone who has ever looked at noisy data knows that if you start looking at derivatives of noisy data the noise is sure to crowd out the data. Frankly, I'm a little disappointed that he would resort to such a cheap bit of fakery - suitable for fooling the bozos I suppose.

If you look at the actual temperatures instead of the time rate of change, you can see something more meaningful and familiar: a noisy signal, with the warmest temperatures heavily concentrated in the late twentieth and early twentyfirst century.



UPDATE: I presented this notion to the Prof and his reply was:

For a random walk, it is extremely likely that the beginning and/or end of an interval maximizes and/or minimizes the function...

Which left me curous as to how likely "extremely likely" was. For a random walk of 350 steps (say of equally likely plus or minus one), just how likely is it that the final step is the maximum of the set? Inquiring statistical minds would like to know. For n=1 I get P=50%, but 1 is a pretty small number...

Comments

Popular posts from this blog

Anti-Libertarian: re-post

Uneasy Lies The Head

Book Review: Anaximander By Carlo Rovelli