Search

Abstract: Receptor models are useful tools to identify sources of a specific pollutant and to estimate the quantitative contributions of each source based on environmental data. This paper reports on similarities and differences in results achieved when testing three receptor models for estimating the sources of polycyclic aromatic hydrocarbons (PAHs) in soils from Huanghuai Plain, China. The three tested models are Principal Component Analysis with Multiple Linear Regression (PCA-MLR), Positive Matrix Factorization (PMF) and Unmix. Overall source contributions as well as modeled ∑PAHs concentrations compared well among models. All three models apportioned three common PAH sources: wood/biomass burning, fossil fuel combustion and traffic emission, which contributed on average 27.7%, 53.0% and 19.3% by PCA-MLR, 36.9%, 27.2% and 16.3% by PMF, and 47.8%, 21.1% and 18.3% by Unmix to the total sum of PAHs (∑PAHs), respectively. Moreover, the spatial evolution of the common sources were well correlated among models (r=0.83–0.99, p <0.001). In addition, the PMF and Unmix models allowed segregating an additional source from the fossil fuel combustion source, with 19.6% and 11.8% contributions to ∑PAHs, respectively. The current findings further validate that different receptor models provide divergent source profiles, which are mainly attributed to both the model itself and/or the underlying dataset. It is therefore generally recommended to apply multiple techniques to determine the source apportionment in order to minimize individual-method weaknesses and thereby to strengthen the conclusion. [Copyright &y& Elsevier]