Comparative receptor modelling study of TSP, PM2 and PM2-10 in Ho Chi MinhCity

Elemental compositions were measured for TSP (total suspended particulate m atter), PM2-10 and PM2 (particulate matter with aerodynamic diameters from 2 to 10 mum and less than 2 mum, respectively) in Ho Chi Minh City. Concent rations of 23 elements and particulate mass (PM) were used for receptor mod elling to identify and quantify aerosol sources using principal component F actor analysis (PCFA). A suite of factors containing similar elements with significant factor loadings were revealed among the factor matrices, thus f acilitating the identification of common sources for different aerosol type s. These sources include vehicular emissions (Br and Zn), coal burning (Se) , industrial processes (Ce, Co, Cr, Pb and Sb), road dust (Al, Ti, V), soil dust (Fe and Th) and biomass burning (K). Marine aerosols (Na and Cl) and mineral fly ash (Sc and La) were revealed only in the PM2-10 model. For TSP , the last Four sources are combined in one factor. The last (9th) factor i n the PM, model, characterised by a high loading from PM and insignificant loadings From elements, was attributed to secondary sulphates and organics, although these constituents were not measured in the experiments. Such a r emarkable source identification capability of the modelling technique highl ights the significance of achieving an optimal factor solution as a crucial step in PCFA, that was done by systematically varying the number of factor s retained and carefully evaluating each factor matrix for both model fitti ng performance and physical reasonableness. (C) 2001 Elsevier Science Ltd. All rights reserved.