<scp>XRF</scp> characterization and source apportionment of <scp>PM10</scp> samples collected in a coastal city

In the current study, a yearlong measurement campaign was conducted during the year 2012 in a medium sized coastal Greek city, Patras. PM10 samples were collected once every 3 days, and a number of those samples were analyzed by the use of a commercial X‐ray fluorescence system, Epsilon 5 by PANalytical, The Netherlands. PMF model was used for source identification. Because the uncertainty of the measurements is used as input in the model, special emphasis was given in its accurate estimation. Seven PM10 emission sources were identified using PMF 5.0 and were, namely, mineral dust (15%), road dust (4.6%), shipping emissions (3.8%), sea salt (11.9%), biomass burning (6.9%), traffic (46.2%), and sulfates (11.6%). The concentration weighted approach was used to investigate if the contributions of the sources identified in the area are affected by long range transportation events. A methodology of estimating the uncertainty of the day to day source contributions is proposed in this study. A comparison between the 24‐hr contribution for the mineral dust factor provided by the model and the calculated contributions for the same factor deduced from appropriate equations (chemical reconstruction) can be used for this purpose. The analysis showed that when the concentrations of the elements associated with the mineral dust source are close to their lowest value, the model assigns zero contribution to the mineral dust source. Following this methodology in the current study, 6 points were identified as bad fitting points, a number that represents less than 10% of the total measurements.