Length of day multifractal dynamics

In this work, the power spectral densities (PSD) of several combined and instrumental length of day (LOD) multifractal time series are computed. As a result, a stable power spectrum, practically independent of the type of time series, of multifractal structure is obtained. The power spectrum of the duration of the day can be divided into three frequency bands. For periods of less than three months, the LOD behaves like a power noise with a score close to $-1.5$. Over periods of three months to three years, LOD behaves almost like white noise. Finally, over periods of more than three years, the LOD also exhibits power noise behavior with a score of about $-1.5$. It turns out that the revealed structure practically does not depend on the type of time series (combined or instrumental), its length, and the parameters of the method for assessing PSD.