Sunday, November 06, 2011

Has the US Been Cooling? (No.)

Anthony Watts tells us that the continental US is cooling on alternate Tuesdays in months that have two full moons.

OK, the cherry-picking isn't quite that bad, but almost. He tells us that Burt Rutan tells him that winters are cooling in the USA-48 at vast rates of up to 8°F per decade! (At that rate we'll disprove the Third Law in about 600 years -- yet again proving the consensus wrong!) And that the whole USA-48 is cooling at -0.87°F/decade for the last 10 years.

People, let's get a grip.

First of all, the USA-48 is only 5.4% of the Earth's land area, and 1.6% of its total area. We aren't the world, as the song (doesn't) go.

Second, a 10-yr time span is meaningless. There are too many natural fluctuations of order 0.2-0.3°C to pick out a manmade trend. Deniers keep ignoring this because the last 10-yrs is about all they have left to bitch about. Santer et al found that at least a 17-yr time interval is needed to pick out an anthropogenic component from global average surface temperature:

Santer, B. D., et al. (2011), “Separating Signal and Noise in Atmospheric Temperature Changes: The Importance of Timescale,” J. Geophys. Res., doi:10.1029/2011JD016263, in press.

Third, other data and other simple calculations don't find these results. I'm using the USA48 data from UAH, because it's what I have handy in a spreadsheet. (It's in the last column.) You can slice the data up in all kinds of ways; here are three:
  • The change in the 10-year moving average over the last 10 years is +0.10°C.
  • The 10-year change in the 12-month moving average is +0.24°C.
  • The slope of the last 10-years/120-months is -0.007 ± 0.008°C/decade. That is, it's not statistically significant.
I'm sure you can cherry-pick something out of the BEST or UAH data that finds cooling for some temperature-like aggregation of some 10 year set of the data, but you need to try rather hard. It will probably help if you first stand on your head.

PS: Notice how they have all suddenly switched to Fahrenheit, because it makes the numbers 1.8 times larger?

No comments: