THE SPECTRUM LENGTH METHOD IN QUANTITATIVE INTERPRETATION OF SELECTED OPTICAL SPECTRA

Abstract

Different algorithms are used for the quantitative interpretation of optical spectra. Regression methods e.g CLS (Classical Least Squares) or PLS (Partial Least Squares) are often used in typical problems of laboratory spectroscopy. The spectra analysis is generally a multi-step process, in which, depending on the spectra type, modification of individual regression methods or special dedicated methods are applied. For example, contour length method or spectra length algorithm are used in this purpose. The basic version of the algorithm is very simple in terms of mathematics. It has been proposed for the analysis of spectra in the OP-FTIR open path spectroscopy, where there are significant fluctuations in the spectrum baseline. The spectrum length is a parameter closely related to the content of the analyzed gas component. Independently of spectral length algorithm, a method has been developed in which difference length of two spectra (measured and reference) is used to compare the spectra. The minimum length of the spectral difference is an indicator of their best fitting. The article presents two ways of using the spectrum length: as a direct and indirect parameter indicating the measured quantity on the basis of the spectrum. There were performed the spectrum length algorithm as a digital differentiating filter with a specific frequency response. There were also analyzed more advanced differentiating filter and the possibility of frequency spectrum filter design used in an analysis of optical spectral signal. There were performed the analysis of the spectrum length method in the case of the synthetic calibration involving the use of model spectra from simulations realized by HITRAN database. There were analyzed numerically the problem of the Instrument Line Shape influence on the synthetic spectra and results of the determining the content of components by the minimizing the difference of spectral length.