Wednesday, July 08, 2009

The phony climate debate over smoothed trends

There are a number of numbers that people compute which appear to be due solely to the fact that with so much computing power at people's fingertips, the urge to calculate cannot be restrained. Thus people calculate for every stock in the stock market its "200-day moving average." The point eludes me. You can't sell for the 200-day average. That 200 days is significant suggests that the dynamics of the stock market are somehow driven by base-10 arithmetic. Why not a 128-day (200 in octal) average?

But I digress. The ability to compute has led climatologists to calculate smoothed trends for climate changes. A well known German named Stefan Rahmstorf has determined that with his secret smoothing technique, a long-term global warming trend is obvious. Steve McIntyre at Climate Audit is arguing that Rahmstorf's method is actually a 15-year triangular filter with some serious defects.

I have a more fundamental point. A 15-year smoothed trend isn't anything real. It's a construct. If it were a construct that helped, it would be one thing. Kinetic energy is defined as one half mass times speed squared. Why not the whole shebang? Because at one half, and with a well chosen definition of potential energy, you get an extremely valuable principle of conservation.

So artificial constructs are not automatically to be discarded, but one has to wonder what value there is in an oddly weighted 15-year average of temperature. Temperature at a given moment is a valuable piece of knowledge. Temperature over a 24-hour period can give a reasonable estimate of the need for heating a home, for instance. There may be some physical significance to the average of a year, but I'm not sure.

But a weighted number over 15 years? It's gibberish. The granularity becomes 15 years. You can do the computation once and then repeat 15 years later, although it's then no clear why you wouldn't just do an equal weight average.

But to get an annual trend out of a 15-year smoothing, as climatologists now do and announce breathlessly to the press, is preposterous. There is no physical reason, no theoretical formula, that depends on the period or the smoothing method. It's a construct without purpose except to impress the ignorant reporters of the mainstream media.