Lines, Sines, and Curve Fittings 17 – More Morlets
I think we can conclude that the ~60 period cycle is strongest in NH Land and weakest in SH Sea, is stronger in the last half of the 20th Century and weakest in the last half of the 19th Century.
There are indications of shorter period cycle in the Southern Hemisphere especially in GISTEMP. Also stronger in the last half of 20th C. Is this a feature of GISTEMP? Or a bug? Is this is a sparsity of data, quality of data issue? Or just the way it is with these quasi-periodic signals.
If I understand this correctly, the hatched regions in the above plots indicate areas where a “cone of influence” reduces confidence due to end-point issues. In other words, the periods examined are long enough to be extending past the end of the series.
I had to hack the wavelet.plot code in the dplR package. Easy enough, since the plot code was written in R itself. The new plot code is here: morlet.plot.R.