Abstract: The least square algorithm using forward and backward linear prediction (LS algorithm)for the autoregressive spectral estimates of the SMMW Fourier transform spectroscopy data is presented. The improvements obtained from the LS algorithm over the maximum entropy algorithm (Burg algorithm) include higher resolution in the spectrum, less bias in the frequency estimate of spectral components, and absence of observed spectral line splitting. The effects of the width of spectral window and the noise on the resolution are discussed.

Key words: Autoregresssive spectral estimates, LS algorithm, Resolution