Monday, June 23, 2008

Principal components analysis (PCA) for source identification

Source identification for atmospheric aerosol is important for developing effective strategy to reduce their emissions. One of the methods for source identification is principal components analysis (PCA). My attention was drawn to this method recently first through a review article about methods and results for aerosol source apportionment over European region and then through the article “Identification of PM sources by principal component analysis (PCA) coupled with wind direction ”. Dr. Viana Rodríguez, Mª del Mar is the lead author on both the articles . She is a researcher at the Institute of Earth Sciences Jaume Almera, Spain.

Application of the principal components analysis method for aerosol is based on the foundation that each source has unique blending of various aerosol components. Variability of the components is strongly correlated among themselves when they are coming from same source when compared to a case where they are coming from heterogeneous sources. Mathematically, PCA seeks to determine matrices A and S in the equation C=A∙S, where column matrix C represents concentration of various particulate matter (PM) components, S is the source contribution and A is the source profiles.

The article Viana et al. (2006) is more about results obtained using PCA analysis rather than the method itself. I liked the article for its clear conclusions and bold figures. I think figure 1 will be useful for those who are interested in knowing typical combination of various aerosol sources for given total mass. Figure 4 will be very useful for the people of the town Llodio, for knowing where to look for reducing pollution. I have recreated images for quick look using data from the manuscript.


Viana, M., et al., (In Press), Source apportionment of particulate matter in europe: A review of methods and results, Journal of Aerosol Science, Accepted Manuscript.

Viana, M., X. Querol, A. Alastuey, J. I. Gil, and M. Menéndez (2006, December),Identification of pm sources by principal component analysis (pca) coupled with wind direction data, Chemosphere 65 (11), 2411-2418.