Signal Processing  Temporal Limitation 

The primary direct spectral estimate from a single data set has a high spectral resolution for the prize of a high estimation variance. Therefore, means for reducing the estimation variance are required, e.g. subdivision of the data set into blocks and averaging the spectra from the shorter blocks. This way real correlations between data at the edges between consecutive blocks are not considered. Alternatively, the highresolution spectrum from the entire data set can be transformed into the appropriate correlation function, the long tails of the correlation function can be clipped and the shorter correlation function can be transformed back into a spectrum via the discrete Fourier transform. The obtained spectrum then has lower spectral resolution at significantly reduced random errors. original literature:
