When is a trend a trend?

I'd like to direct your attention to the whirlwind of debate over an 11 Jan post at realclimate.

The hullabaloo seems to be over use of weather trends in 7 year data sets. Long story short, seven years appears far to short to be used as an indicator global of warming or cooling. Longer is better and 15 years seems to be a good indicator of warming / cooling trends.

It's disheartening to see that a basic concept like this* is the source of such controversy in the present day. When confusion is easily sown amongst people who ought to know better, then there's clearly a backstory that I could investigate, but why should I involve myself with such a politicized crapfest?

The "basic concept" is that any global climate trend is a small delta per year and is overcome by the larger (normal, intrinsic) global weather variation. This issue has analogs in many scientific fields, but is usually easily overcome with statistics: The normal response in other scientific fields to this problem is to make many parallel or sequential observations of identical systems. Multiple observations improve the accuracy with which one can observe a trend by the square root of the number of times you make the measurement (e.g. 9 measurements means 3 times the accuracy). Climatologists can't do this for obvious reasons, so they are stuck with high uncertainty and a low data acquisition rate: the data are noisy and sparse.

No comments:

Post a Comment