Post submitted by John A.
Readers of Climate Audit will have noted Steve McIntyre mentioning Chladni patterns in respect to the results of some climate reconstructions which exhibit spacial autocorrelation – which means that the data points are related to each other depending on how far away from each other they are, and thus by the shape of the area which contains the data points.
The notable example being Steig, Mann et al 2010 which purported to reconstruct temperatures in Antarctica and demonstrate a warming of West Antarctica. That study turned out to be caused by the poor modelling and worse statistical analysis, as shown by O’Donnell et al 2011. (Who would have expected a piece of statistical analysis to be so flawed when done by Michael Mann? I was as shocked as the rest of you).
Steve noted (but it didn’t survive the Steig-acting-as-a-conflicted-peer-reviewer cut) that the eigenvector patterns that were supposed to reflect temperature changes were in fact patterns which arise from autocorrelated noise.
The Chladni patterns can be shown easily using a square metal plate which has rice scattered on it, sitting on top of an amplifier. The amplifier sends a pure frequency sound to the metal plate, and in the demonstration below, that frequency is slowly increased. The patterns change dramatically as the signal frequency is slowly changed.
I think its a cool video. Enjoy.