Positive Matrix Factorization dynamics in fingerprinting: A comparative study of PMF2 and EPA-PMF3 for source apportionment of sediment polychlorinated biphenyls

Karakas F., İmamoğlu İ. , Gedik K.

ENVIRONMENTAL POLLUTION, vol.220, pp.20-28, 2017 (Journal Indexed in SCI) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 220
  • Publication Date: 2017
  • Doi Number: 10.1016/j.envpol.2016.07.066
  • Page Numbers: pp.20-28
  • Keywords: PCBs, Environmental fate and transport, Aquatic environment, Receptor model, PMF 5.0, FINE PARTICLES, DECHLORINATION, POLLUTANTS, MODEL


Receptor models were typically used in air pollution studies and few publications are available for Positive Matrix Factorization (PMF) that consider the details of parameters and procedures in evaluating the trace organic pollutants in sediments. In this study, environmental fate and source composition of Lake Eymir sediments contaminated by polychlorinated biphenyls (PCBs) were explored by applying two PMF models, Paatero's PMF2 and United States Environmental Protection Agency's (US EPA) EPA-PMF3. PMF2 and EPA-PMF3 rely on different algorithms; Paatero's algorithm and multilinear engine algorithm, respectively. Here, the approaches of two PMF models were compared for the identification of PCB patterns taking into consideration the effects of various uncertainty matrices, residual matrices and goodness-of fit parameters. As a result of the study, it was understood that both models resolved five factors and indicated Clophen A60 as the source of PCBs. These results were consistent with the results resolved by Chemical Mass Balance model applied to the same data set in a previous study. However, source contributions identified by two models differed in quantity, but with similar patterns. This study indicates a way in understanding behavior, fate and global source of persistent organic pollutants in sediment by applying and comparing with a special data including high percentage of below detected value (38.2%) to understand the dynamics of PMF model parameters. (C) 2016 Published by Elsevier Ltd.