Source Apportionment of Aerosol in Delhi using UNMIX Receptor Model

Abstract

PM1 (particles having aerodynamic diameter < 1.0 μm) concentrations were measured in previous study at a sampling site inside the Indian Institute of Technology (IIT) Delhi campus for 5 months from November- 2009 to March- 2010. In this study the source identification and quantification were done using UNMIX (Version 6) model. The input data included 15 species ( Na, K, Cl, Mn, SO4, NO3, Ca, NH4Cd, Cr, Cu, Ni, Zn, Pb, Fe, and Mg) for 51 days. The sources with eigenvalue >1 were finally selected by model as clear source and other with eigenvalue<1 were omitted by default. UNMIX provided the results in three different ways namely Analyze Run, Highlight Run Output, or Diagnostic Plots. The three sources were identified by model as clear sources. For source profile 1 composition analysis showed high contribution from Mn, and Ca, but associated with large uncertainty at 95% CI. The factor loading analysis also revealed high loading of Ca (0.63), Mn (0.85) to source 1. The source profile 1 was partially contaminated with NH4 and SO4 as moderate to low contribution with low uncertainty at 95% CI was estimated. One of the plausible explanation could be the crustal soil particle were enveloped by sulphate aerosol forming shell around the dust core. The above analysis revealed that source profile 1 was primarily influence by Ca and Mn and can be interpreted as crustal soil. In addition to this the EF Crust values of individual elements in PM1 size fractions were analysed, which varied 2-9 for Ca and 4-8 for Mn in different months. The EF crust less than 10 indicates these elements are predominantly from crustal soil confirming the model interpretation. Source 2 highly loaded with elements K (0.84) and Na (0.90) can be interpreted as biomass burning. The evidence of open agricultural field burning and biomass fuel burning for space heating during winter have been reported in literature (Parmer et al., 2001; Nair et al., 2006, Chowdhury et al., 2007) for peninsular and Delhi region. Source 2 was validated using the K/Na ratio for different biomass burning reported in literature. The K/Na ratio varies 0.6 to 2.92 for various type of biomass burned. The ratio estimated in present study also varied from 0.98 to 1.86 lies in the range of values reported in literature indicting the strong influence of biomass burning during winter in Delhi region. Sources profile 3 was significantly influenced by SO4-2, NO3- and NH4+ with large uncertainty, however, again the fraction apportionment of these species to source profile 3 indicated low variability. In factor loading analysis source 3 was significantly loaded with IV NO3- (0.88), SO4-2 (0.84) and NH4+ (0.89) indicating the secondary aerosol formation mechanism as a source. The formation of SO4-2 and NO3- in the atmosphere takes place by photochemical reaction (Chakroburty and Gupta, 2010, Saolapurkar and Sharma, 2006), and are termed as secondary aerosol. Fog during winter the during winter the photochemical reactions of precursor gases (SO2 and NOx) in the presence of UV radiation result in aerosol formation. Singhai (2010) reported high concentration of SO4 ions in densely foggy days of December, 2009 and January, 2010. Aerosol particles mediate the formation of fog in the atmosphere (Pandis et al., 1990; Seinfeld and Pandis,1998) through the preferred heterogeneous nucleation of water vapour on pollution particles, at high RH. Fog droplets further aid aerosol formation through aqueous-phase reactions of soluble gaseous precursors (e.g. SO2 and H2O2), leading to higher aerosol concentrations of species like sulfate on fog abatement, which then nucleate subsequent fog–smog–fog cycles (Pandis et al., 1990). Particulate inorganic matter would also contain soluble inorganic ions (Cl_, NO3-, NH4+) present in combustion aerosols, which could also mediate water uptake and fog formation (Mehta et al., 2009) hence, again enhance the sulphate aerosol formation. The secondary aerosol formation is the major contributor (67%) to atmospheric aerosol loading over Delhi region followed by crustal soil (22%) and biomass burning (10%). The finding in this study has implications in regional climate study. The model predicted element concentrations were correlated with observed concentration. The predicted Vs measured PM1 concentrations plot (r2=0.77) showed that the UNMIX model worked well and the predicted and measured PM1 mass concentrations are in good agreement. The NO3-, K, NH4+ SO4-2 and Ca showed high correlation with r2 value greater than 0.7, while Cl, Na, and Mn showed moderate correlation with R2 varied between 0.52- 0.67.