Fluxes of galactic cosmic rays (GCR) observed at 1 AU are modulated inside the heliosphere at different time scales. Here we study the properties of the power spectral density (PSD) of galactic cosmic ray variability using hourly data from 31 neutron monitors (NM) from 1953 to 2016. We pay particular attention to the reliability of the used datasets and methods. We present the overall PSD and discuss different parts of the spectrum and the related periodicities. We find significant spectral peaks at the periods of 11 years, 1.75 years, 155 days, 27 days and 24 hours and the harmonics of the latter two peaks. We calculate a power law slope of −1.79 ± 0.13 for the period range between 50 and 130 hours and a slope of −1.34 ± 0.17 for the period range between 40 days and 3.4 years (1000 − 30000 h).

We use the principal component analysis to study geomagnetic activity at annual resolution using a network of 26 magnetic stations in 1966-2015, and an extended network of 40 stations in 1980-2015. The first principal component describes the long-term evolution of global geomagnetic activity, and has an excellent correlation with indices like the Kp/Ap index. The two networks give identical results for the first principal component. The second principal component is highly correlated with the annual percentage of high-speed streams (HSS). The extended network has a slightly higher sensitivity to HSSs. We verify the non-trivial latitudinal distribution of the second empirical orthogonal function. We find that the amplitude of the 22-year variation of geomagnetic activity has a closely similar latitudinal distribution as the second empirical orthogonal function. This verifies that the 22-year variation of geomagnetic activity is related to HSSs. The most likely cause is the Russell-McPherron mechanism.