Premium
Synergistic use of several receptor models (CMB, APCS and PMF) to interpret air quality data
Author(s) -
Andriani E.,
Caselli M.,
de Gennaro G.,
Giove A.,
Tortorella C.
Publication year - 2011
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.1120
Subject(s) - principal component analysis , multivariate statistics , particulates , environmental science , apportionment , air quality index , aerosol , environmental chemistry , atmospheric sciences , statistics , chemistry , mathematics , meteorology , physics , organic chemistry , political science , law
PM 10 samples were collected in Civitavecchia (Rome, Italy) during a monitoring campaign performed between August and September 2008; elemental, ionic and organic characterisation of the samples was carried out. Three receptor models (Chemical Mass Balance, Absolute Principal Component Scores and Positive Matrix Factorisation) were applied to identify PM 10 sources and their contributions. The results obtained by the models were compared in order to perform a robust characterisation of the sources. Seven emission sources were considered relevant for the area under investigation and their profiles were used to run CMB. Source percentage contributions to the total particulate matter samples were evaluated: the greater contribution was obtained for Traffic (39%), the lower for Marine Aerosol (3%). Five factors were selected to perform the multivariate receptor models (APCS and PMF) after the application of explorative Principal Component Analysis to the dataset. Furthermore, attention was focused on the characterisation of source profiles and contributions obtained by PMF because it best reconstructed the PM mass of the samples. The percentage contributions of the five sources to the total PM samples were estimated: the greater contribution was obtained for resuspended matter (22%), the lower for combustions (17%). The results of the source apportionment studies can be used by air quality managers to develop appropriate control strategy. However, the differences in sources profiles and weights obtained by the models were related to their specific approaches: further investigations are suggested in order to develop sources profiles more reliable for the area under investigation. Copyright © 2011 John Wiley & Sons, Ltd.