Assessment of factors influencing PM mass concentration measured by gravimetric & beta attenuation techniques at a suburban site

Abstract Near real-time atmospheric particulate matter (PM) monitors are extensively used in air quality networks given their ability to provide continuous measurements with minimal attention by the operator. Their principle of operation is based on measurement of a physical parameter that is quantitatively linked to the PM mass concentration. Significant discrepancies between these measurements and those obtained by the reference gravimetric method, conducted in regions with diverse climatic conditions, have been reported in the literature. In this study we compare systematic PM2.5 and PM10 gravimetric (GM) and beta attenuation (BA) measurements performed at a suburban site in Athens, Greece, over a period of 4 years (2009–2012). In general, BA and GM datasets exhibited similar temporal variation for both PM size fractions. An overestimation of the ΒΑ measurements, which was ∼30% for the PM2.5 and ∼10% for the PM10 data, was observed. Good linear correlations between GM and BA data were observed, with estimated Pearson coefficients being 0.79 for the PM2.5 and 0.85 for the PM10 measurements. The respective fitted equations through the entire dataset were BA = 0.71 GM + 6.2, and BA = 0.77 G M + 4.1. Better correlation between GM and BA measurements was observed during the cold rather than the warm period. Discrepancies between BA and GM PM2.5 measurements increased with increasing available water vapor, suggesting that the aerosol bound water has a strong effect on the measurements. The effect of filter material used for GM measurements (i.e., quartz, glass fiber, or Teflon) was also examined for the PM2.5 dataset. Best correlation between BA and GM data was observed when glass fiber, which is incidentally the material of the BA filter tape, was used in the GM measurements. When the BA to GM relationship was examined by further categorizing the data by the season (i.e., cold and warm period) for different filter types, the relationships that were fitted to the data for the two seasons were similar when Teflon filters were used, but quite diverse when quartz and glass fiber filters were employed in the GM measurements. Finally, the variability of the ratio between the two measurement techniques was found to be potentially dependent on the availability of the volatility or stability in the aerosol phase of species such as ammonium nitrate.