Classification of aerosol LAMS mass spectra using discriminant analysis

The identification of ambient particulate matter emission sources was the objective of this study. Discriminant analysis (DA) was applied to aerosol mass spectra obtained by a laser ablation mass spectrometer (LAMS). The spectral data was obtained between January 2002 and February 2003. Data collected in on an hourly basis was used to determine the classification variables. Statistical and mathematical packages calculated the classification functions and discriminant scores. Two databases were used to determine the feasibility of discriminant analysis (DA) to classify aerosol mass spectra into nine pre-defined sources. The accuracy of DA to classify data correctly was determined by using a dataset that was previously classified by positive matrix factorization (PMF). The acquired results demonstrated the potential of the DA for source identification.