Barents Sea Ice edge position 1800-2018. The red line is the parabolic fit
used for Lomb spectrum. Data are the previous ones.
Yndestad and other collegues, specialists of the same argument (the 18.6 yr
period of lunar -nodal, Saros- influence), like so much to interpolate data
(they need constant step in wavelets analysis) and to filter in order to
avoid noise. I don't agree with such a position and use data "as measured"
without inventing (interpolation) non existent data or cancel some of them (
which someone define "noise" when in reality they are possible "signal".
The spectrum shows a low-power 18.4 yr spectral peak and very little of the
*plaetora* of sub- and super-harmonics listed by Yndestad (2019, blog page)
and often given with 3 or more decimal digits. I can find the harmonics with
periods similar to the theoretical ones, without no warranty of correctness
(e.g. 75.9 yr instead of 74.4 yr: are they the same? And at what degree of
confidence? And, again, in which way can I correcly compute such a confidence?).
I need to always remember my data are digitized from figure 7 of Yndestad's blog
post.