Partic. vol. 37 pp. 107-118 (April 2018)
doi: 10.1016/j.partic.2017.05.009

Source apportionment of PM₁₀ in Delhi, India using PCA/APCS, UNMIX and PMF

Srishti Jain^a,b, S.K. Sharma^a,b,*, T.K. Mandal^a,b, Mohit Saxena^a

sudhir.npl@nic.in

Abstract

Source apportionment of particulate matter (PM₁₀) measurements taken in Delhi, India between January 2013 and June 2014 was carried out using two receptor models, principal component analysis with absolute principal component scores (PCA/APCS) and UNMIX. The results were compared with previous estimates generated using the positive matrix factorization (PMF) receptor model to investigate each model’s source-apportioning capability. All models used the PM₁₀ chemical composition (organic carbon (OC), elemental carbon (EC), water soluble inorganic ions (WSIC), and trace elements) for source apportionment. The average PM₁₀ concentration during the study period was 249.7 ± 103.9 μg/m³ (range: 61.4–584.8 μg/m³). The UNMIX model resolved five sources (soil dust (SD), vehicular emissions (VE), secondary aerosols (SA), a mixed source of biomass burning (BB) and sea salt (SS), and industrial emissions (IE)). The PCA/APCS model also resolved five sources, two of which also included mixed sources (SD, VE, SD+SS, (SA+BB+SS) and IE). The PMF analysis differentiated seven individual sources (SD, VE, SA, BB, SS, IE, and fossil fuel combustion (FFC)). All models identified the main sources contributing to PM₁₀ emissions and reconfirmed that VE, SA, BB, and SD were the dominant contributors in Delhi.

Graphical abstract

Keywords

Receptor model; PCA/APCS; UNMIX; PMF; Source apportionment