Combined method for detecting hidden anomalies in galactic cosmic ray variations

A method has been proposed for detecting hidden periodicities in nonstationary finite time series with strong noise. The successive application of the spectral singular expansion and wavelet transform makes it possible to calculate the energy of short signals with an arbitrary form against a suppressing noise back-ground, the amplitude of which is several times as large as that of a useful signal. This method has been used to identify variations in the galactic cosmic ray (GCR) flux related to IMF disturbances due to the earthward propagation of solar magnetic clouds in the heliosphere.