Geophysical processes and biosphere: article

Аnnotation: Until now, the main carrier of information in biological and heliogeophysical signals was the sum of periodic variations. However, not all heliobiological time series contain dominant periodic components. Relaxation analysis in this paper allows us to generalize the «spectral» formalism to signals that cannot be represented as a sum of a limited number of quasi-periodic components. The algorithm for filtering noise and long-period trends is developed, based on the separation of the original signal into fast and slowly relaxing components (applicable not only to one-dimensional series, but also to two-dimensional images), and the main theorem that guarantees its operability is proved. A method for constructing an orthonormal basis whose components have a strictly defined relaxation time is described. The result of signal decomposition over this basis is called the relaxation spectrum. With its help, the time series can be divided into «signal-to-noise» or «oscillation-trend» in the absence of adequate Fourier or stochastic models. Using an automatic extraction algorithm we verified spectral peaks of human population and heliogeomagnetic time series and identified similarity of these peaks through Sørensen’ measure. Our results confirmed earlier hypothesis in heliobiology that geomagnetic rhythms with periods of near 7, and 9 days most significantly affect physiological parameters of biological objects (population in general).