Numerical Pattern Recognition Analysis of CO Atmospheric Simulation Experiments

Abstract

A technique entitled Hybrid Linear Pattern Analysis (HLPA), which represents a combination of model-based and pattern recognition-based approaches to the analysis of spectroscopic data, is introduced and applied to the analysis of time-resolved infrared emission spectra of ground electronic state (X1Σ+) CO obtained in atmospheric simulation experiments. The spectra are highly congested and consist of incompletely resolved, overlapping v‘ − v‘ ‘ = 1 emission bands from v‘ = 1 up to at least v‘ = 12. The analysis of the time dependence of the emission intensity in the various vibrational bands had been stymied by a severe optical opacity effect in the v = 1 → 0 emission, which is difficult to simulate; thus, conventional least-squares fitting could not be used confidently to determine the time-dependent emission intensity of this band, or that of at least three other emission bands that overlap strongly with it. The HLPA technique permits an alternate approach in which the v = 1 → 0 emission band is considered to be an unknown pattern that is identified by the Extended Cross Correlation (XCC) pattern recognition technique (J. Chem. Phys. 1997, 107, 8349). The intensity profiles of the other bands, however, can be predicted accurately based on the experimental parameters, and this knowledge is used in conjunction with the results of the XCC to determine the time dependence of all of the vibrational bands, and the intensity profile of the v = 1 → 0 emission band.