Estimation of spectral moments of time series

SUMMARY The task of estimating moments of the spectrum of a stationary time series, rather than the more common problem of estimating the spectrum itself, is discussed. In particular, explicit formulae are deduced for the mean and variance of an estimator of the normalized standard deviation of the spectrum. Besides its theoretical significance, the necessity for estimating such a moment is of practical importance, for example, in the analysis of the velocity distributions of turbulent phenomena.