Spectral analysis of sunspot number and geomagnetic indices (1868-2001)

The multiple-taper spectral analysis was applied to a(a), A(p) and R-z annual and monthly average series (1868-2001). The significant periods at 95% confidence level for annual averages are 11.1 and 10.2 years for R, and 11.1, 5.3, 4.3, 2.7 and 2.1 years for a(a). For monthly averages, the significant periods were 133 (11.1 years) and 17.7 months for R. and 133 (11.1 years), 63.7 (5.3 years), 17.5, 12.7 and 6 months for aa. The periods associated with the solar cycle are observed in both series. The aa spectrum showed in addition a peak near 4.3 years, which is not present in the R-z spectrum. Other works have also shown this peak in geomagnetic activity. This peak is believed to be caused by the dual-peak structure in the a(a). The 22-year Hale cycle was not detected either in R-z or a(a) spectrum in the period 1868-2001, but it was seen in R-z data for 1700-2000 annual averages. The explanation of this 22-year cycle has been questioned and while in some spectral analysis this periodicity has been found, others have failed in its detection. The aa spectrum showed in addition a peak of 6 months, associated to the semi-annual variation of geomagnetic activity. Three hypotheses have been considered to explain the seasonal activity variation. These are the Russell-McPherron effect, the equinoctial effect and the axial effect. (c) 2005 Elsevier Ltd. All rights reserved.