Applying receptor models Unmix and PMF on real data set of elements in PM for sources evaluation of the sea coastal side region (Southeast Adriatic Sea)

Abstract. Two receptor modeling techniques were applied to a common data set of daily measurements of 11 elements in particulate matter (PM). Samples of PM were collected in the 5 yr period at an urban site located at the sea coast. In the vicinity of the sampling site traffic is a permanent but not significant anthropogenic source. In this study we used both the Unmix and PMF receptor models for evaluation of the sources contribution. Unmix found thirteen solutions for several combinations of species, but four solutions satisfy the criteria Min R² > 0.8 and Min S/N > 2. Unmix identified three and four sources in satisfactory solutions. The PMF model has given 3 possible solutions and by further analysis the best solution of four sources was selected. F peak refinement enabled finding a more realistic solution that includes re-suspension and traffic as dominant source contributions. The results given in this study are in accordance with the results of Enrichment Factors analysis presented in our previous work.