THE DETERMINATION OF CONFIDENCE-LIMITS ASSOCIATED WITH ESTIMATES OF THE SPECTRAL PEAK FREQUENCY

A Monte-Carlo simulation technique is used to evaluate the suitability of proposed techniques for the estimation of the spectral peak freque ncy. Due to the statistical variability of spectral estimates and the discrete frequency resolution of spectra, the calculated values of pea k frequency are stochastic variables. The probability density function s for such estimates are functions of the frequency resolution and the number of degrees of freedom of the spectrum from which they are deri ved as well as the spectral peakedness. The mean values of all techniq ues are biased high, indicating the derived values of peak frequency a re an overestimate of the true value. The probability density function s do not follow an obvious analytical form. Tabular values are, howeve r, presented to enable the determination of confidence limits for esti mates of the peak frequency.