A Comparison of Four Gravimetric Fine Particle Sampling Methods

• Published in: Journal of the Air & Waste Management Association (1995) Vol. 51; no. 6; pp. 878 - 884
• Main Authors: Yanosky, Jeff D., Maclntosh, David L.
• Format: Journal Article
• Language: English
• Published: Pittsburgh, PA Taylor & Francis Group 01.06.2001; Air & Waste Management Association
• Subjects: Air Pollution - analysis; Air Pollution, Indoor - analysis; Analysis methods; Applied sciences; Atmospheric pollution; Environmental Monitoring - instrumentation; Environmental Monitoring - methods; Exact sciences and technology; Gravitation; Particle Size; Pollution; Reproducibility of Results; Sensitivity and Specificity; Human; Gravimetric analysis; Data analysis; Statistical analysis; Personal sampling; Air quality; Exposure; Regression analysis; Suspended particle; Ambient air concentration; Fine particle; Content; Sampler; Aerosols; Indoor pollution; Air pollution; Measurement accuracy; Sampling; Performance; Measurement error; Comparative study
• ISSN: 1096-2247; 2162-2906
• DOI: 10.1080/10473289.2001.10464320

A study was conducted to compare four gravimetric methods of measuring fine particle (PM 2.5 ) concentrations in air: the BGI, Inc. PQ200 Federal Reference Method PM 2.5 (FRM) sampler; the Harvard-Marple Impactor (HI); the BGI, Inc. GK2.05 KTL Respirable/Thoracic Cyclone (KTL); and the AirMetrics MiniVol (MiniVol). Pairs of FRM, HI, and KTL samplers and one MiniVol sampler were collocated and 24-hr integrated PM 2.5 samples were collected on 21 days from January 6 through April 9, 2000. The mean and standard deviation of PM 2.5 levels from the FRM samplers were 13.6 and 6.8 μg/m 3 , respectively. Significant systematic bias was found between mean concentrations from the FRM and the MiniVol (1.14 μg/m 3 , p = 0.0007), the HI and the MiniVol (0.85 μg/m 3 , p = 0.0048), and the KTL and the MiniVol (1.23 μg/m 3 , p = 0.0078) according to paired t test analyses. Linear regression on all pairwise combinations of the sampler types was used to evaluate measurements made by the samplers. None of the regression intercepts was significantly different from 0, and only two of the regression slopes were significantly different from 1, that for the FRM and the MiniVol [β 1 = 0.91, 95% CI (0.83-0.99)] and that for the KTL and the MiniVol [ = 0.88, 95% CI (0.78-0.98)]. Regression R 2 terms were 0.96 or greater between all pairs of samplers, and regression root mean square error terms (RMSE) were 1.65 μg/m 3 or less. These results suggest that the MiniVol will underestimate measurements made by the FRM, the HI, and the KTL by an amount proportional to PM 2.5 concentration. Nonetheless, these results indicate that all of the sampler types are comparable if ~10% variation on the mean levels and on individual measurement levels is considered acceptable and the actual concentration is within the range of this study (5-35 μg/m 3 ).