Figure 1: Statistical complexity versus the (log of) length of the analysis interval for the hydrogen velocity time series (top curve) and four surrogate time series (bottom curves): three independent realisations of the phase-shuffling algorithm (red, green, and blue), and single time series of a white noise passed through a low-pass linear filter (black) (from [3]). Note that the value of statistical complexity for the data obtained from simple random number generator is close to zero and does not depend on the length of time series for large enough value of the latter (not shown).