Improved Detection Limits in Fourier Transform Spectroscopy from a Maximum Entropy Approach to Baseline Estimation

Abstract

A new method for baseline estimation and correction in gas phase Fourier transform (FT) absorption spectra based on a maximum entropy approach is described. In this novel method, a smoothly varying baseline function is determined directly from the observed spectra by separating baseline and signal information on the basis of line width. Baselines of arbitrary complexity may be determined provided there is a difference of at least a factor of 5 in line width between the signal and baseline features and there are no discontinuities in the baseline. As baseline drift is often much larger than the noise in FT spectroscopy, significantly improved detection limits are obtained using this method for the quantitative analysis of trace gases. Even in the absence of baseline drift, the use of an accurate noise-free baseline function will always lead to some improvement in the detection limits. The quantitative results are shown to be insensitive to input parameters over a wide range.