A new spectral analysis method, "Sompi," proposed for seismogram analysis based on the autoregressive (AR) model is shown to be useful in extracting unbiased NMR spectra with high resolution and high sensitivity from FID data. The Sompi method deconvolutes the given time series data
into a set of spectral lines characterized by frequency, decay constant, initial amplitude, and initial phase. In order to determine the number of spectral lines in the AR modeling of FID data, the Akaike information criterion (AIC) is extended to the model with two parameters: an AR order
which specifies the number of the candidate spectral lines, and the number of spectral lines with finite power that exist. The use of the Sompi method combined with the two-parameter AIC is shown to give an accurate and reliable spectrum by analyzing the synthetic data with white noise superimposed.
The spectral resolution by the Sompi method is shown to be improved further by means of an intentional aliasing of narrow-band filtered data. The present method is demonstrated by analyzing the observed NMR data on polyvinylchloride to resolve fine structure which cannot be recognized by FFT.
JEOL Ltd., 1-2 Musashino 3-chome, Akishima, Tokyo 196, Japan 2:
Geophysical Institute, The University of Tokyo, Bunkyo, Tokyo 113, Japan
Publication date: May 1, 1990
More about this publication?
The Society publishes the internationally recognized, peer reviewed journal, Applied Spectroscopy, which is available both in print and online. Subscriptions are included with membership or can be purchased by institutional or corporate organizations. Abstracts may be viewed free of charge. Previously published as Bulletin (Society for Applied Spectroscopy)