Abstract
Pollution source analysis is an effective method that can help chemical park managers accurately understand the characteristics and contributions of pollution sources in the park. However, as more receptor models are being used in this field, it has become difficult for managers to find the best interpretable and reasonable model among many source analysis models. Here, we present a case study of pollution source analysis in a southern chemical park using the D-S evidence theory approach to combine the source analysis results of different receptor models for comprehensive consideration. Receptor models were used to analyse the pollution sources via positive definite matrix decomposition, principal component analysis-multiple linear regression, and Unmix models. The results demonstrated that source analysis was greatly influenced by the uniqueness of pollutant characteristics and model receptor differences. Furthermore, incomparable analysis results and low fineness were observed. The D-S evidence theory model proposed in this study solved the above-mentioned problem to some extent and successfully extracted the four primary pollution sources in the study area, of which 45.73 % came from the metal processing industry (F1), whose primary pollutants were Cr, Ni, Zn, Cr(VI), and Cu, and 25.12 % came from the electronics manufacturing industry (F2), whose primary pollutants were Pb, Cr(VI), Cu, and Zn. 15.62 % of the contamination came from the production of chemical agents (F3), whose primary pollutant was TEHP, and 13.53 % came from the use of oil-containing auxiliary materials (F4), whose primary pollutant was TPH. The D-S evidence theory model used in this study provides a reference for the management of chemical parks.