Tuesday, December 11, 2012

The Closing of the David Rose Hole

Ever since David Rose's article in October claiming that global warming stopped 16 years ago, you hear a lot of 'skeptics' repeating the claim. It is very thin gruel.

As I showed here, it's a total cherry pick, and not even a very good one. There has been warming, it's just not quite significant at the 95% confidence level.

There is more HadCRUT4 data now, and this small hole of no warming is getting smaller. (GISS shows statistically significant warming prior to 2000, so I'll concentrate on HadCRUT4.) There is now only a 6 months period where the warming is not statistically significant at the 95% level:


How small is the David Rose hole? HadCRUT4's warming is only, at worst, 92% statistically significant, instead of the canonical 95%:


This tiny difference is all that the claim now rests on. It will probably be done in a few months anyway.

This (obviously) changes if you include autocorrelation -- the fact that the climate system has inertia, and one month's temperature depends on the temperature of the months before it -- but not by much. Here is the graph of the linear trend including rank-1 autocorrelation (that is, a correlation between immediately neighboring months only):


And the plot of statistical significance:


The math reveals what the physics already knows: 16 years is too short of a time interval to make statistically significant conclusions about temperature trends -- there is just too much autocorrelation (inertia) in the system.

No comments: