Premium
Luni‐solar and solar cycle signals in lake Saki varves and further experiments
Author(s) -
Currie Robert G.
Publication year - 1995
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.3370150805
Subject(s) - varve , series (stratigraphy) , nyquist–shannon sampling theorem , solar cycle , noise (video) , amplitude , variance (accounting) , environmental science , confidence interval , spectral density , statistics , climatology , mathematics , geology , physics , sediment , paleontology , computer science , mathematical analysis , accounting , quantum mechanics , artificial intelligence , solar wind , magnetic field , business , image (mathematics)
Spectrum analysis of over 4000 years of Lake Saki varve data in 200‐year swaths yields evidence for two peaks with periods 18·5±1·5 years (18 out of 21 series) and 10·7±0·8 years (in 15 series). Statistical tests show that the long‐period peak is statistically significant at a 98 per cent confidence level. The peaks are identified as the luni‐solar 18·6‐year, M n , and the 10–11‐year solar cycle, S c , signals in climate. Amplitudes of both terms are highly non‐stationary with respect to time, and the phase of M n wavetrains is also highly non‐stationary, often displaying abrupt changes of 180°. On average the two terms account for 20 per cent of total variance in the raw Saki series, the Nyquist frequency of which is 0·5 cpy. This of course is highly unrealistic—and when the power in the raw Saki data from 0·125 cpy (period of 8 years) to the Nyquist is filtered out prior to analysis, and the variance contribution of the signals to filtered data compared over a common frequency interval (periods 30 to 8 years), their variance contribution increases dramatically to 78 per cent. Such experiments have been carried out for thousands of climate records with similar results, and examples from both published and unpublished data are given herein. Thus, the spectrum of climate from 30 to 8 years has turned out to be strongly ‘signal‐like’ rather than ‘noise‐like’ as radically assumed by statisticians/mathematicians over the past 70 years.