Source apportionment of heavy metals in soil of Guangzhou: Comparison of three receptor models

Receptor models are useful tools to identify the types of pollution source and estimate the contributions of each source of the observed samples. To analyze the concentrations, distributions and sources of eight heavy metals including lead (Pb), cadmium (Cd), zinc (Zn), mercury (Hg), arsenic (As), copper (Cu), chromium (Cr), and nickel (Ni) in soils, 208 topsoil samples were collected in the main urban area of Guangzhou, China. Three receptor models (Multi-Linear Regression of the Absolute Principal Component Scores (APCS-MLR) method, Positive Matrix Factorization (PMF) method and UNMIX method) were employed to identify the potential pollution sources of heavy metals and to apportion the pollution sources. Results show that the mean concentrations of eight heavy metal elements are higher than the corresponding background values, with the mean concentration of Cd being almost five times its background value. The three receptor models all identify three potential pollution sources, which are nature source, traffic source and industry source. Moreover, PMF and UNMIX can identify an agricultural source besides the three pollution sources, which better distinguishes the different types of pollution sources. Comparison among the results of APCS-MLR, PMF and UNMIX shows that there are some significant differences in the estimated contributions for each potential pollution source. It is also found that PMF appears to be more plausible for this investigation. It is advisable to use multiple receptor models to perform source identification and source apportionment, and the results could be very useful to local administrations for the control and management of pollution and better protection of important soil quality.