Principal and independent components analysis of overlapping spectra in the context of multichannel time-resolved absorption spectroscopy
The separation of overlapping absorption spectra in the context of multichannel time-resolved absorption spectroscopy and chemical kinetics is a particular case in the general problem of splitting the observed data into several linear components. Here, principal and independent components analysis a...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2004 |
| País: | España |
| Institución: | Consejo Superior de Investigaciones Científicas (CSIC) |
| Repositorio: | DIGITAL.CSIC. Repositorio Institucional del CSIC |
| OAI Identifier: | oai:digital.csic.es:10261/343459 |
| Acceso en línea: | http://hdl.handle.net/10261/343459 |
| Access Level: | acceso abierto |
| Palabra clave: | Multichannel time-resolved absorption spectroscopy Chemistry of ozone and iodine Overlapping molecular spectra Principal components analysis Independent components analysis |
| Sumario: | The separation of overlapping absorption spectra in the context of multichannel time-resolved absorption spectroscopy and chemical kinetics is a particular case in the general problem of splitting the observed data into several linear components. Here, principal and independent components analysis are applied to kinetic data of iodine-ozone chemistry, which contains overlapping spectra of different absorbers. The objective of this work is to demonstrate a method which in spite of this overlap is able to extract separated time traces of such absorbers. These time traces are clearly a pre-requisite for any further accurate quantitative analysis. The statistical properties of data recordings obtained from flash photolysis of I-2 and O-3 have been studied to check if the requirements of the model are fulfilled. Results of separation in appropriate spectral windows displaying overlapped vibrational features are presented. Validation is made using prior information and conventional techniques. (C) 2003 Elsevier B.V. All rights reserved. |
|---|