How do outdoor particulate matter concentrations measured using Ultrasonic Personal Air Samplers (UPAS) compare to those measured using United States Environmental Protection Agency (U.S. EPA) federal reference and equivalent methods (FRMs and FEMs)?  Keep reading to see the answers we got by collocating UPAS with a Rupprecht & Patashnick Partisol^®-Plus Model 2025 Sequential Air Sampler and a GRIMM Environmental Dust Monitor (EDM) 180 at a regulatory monitoring site in Denver, Colorado!

Collocation at regulatory monitoring site in Denver

From September 3-17, 2020, several UPAS in custom weatherproof enclosures were collocated with a Rupprecht & Patashnick Partisol^®-Plus Model 2025 Sequential Air Sampler and a GRIMM EDM 180 at the Continuous Air Monitoring Program (CAMP) site operated by the Colorado Department of Public Health and Environment (CDPHE) at 2105 N Broadway in Denver, Colorado (AQS site ID 08-031-0002).  The R & P Model 2025, which is a Federal Reference Method (FRM) sampler, collected a 24-h PM_2.5 filter sample every day. The EDM 180, which is a Federal Equivalent Method (FEM) for PM_2.5, measured PM_2.5 and PM₁₀ concentrations continuously.

Each UPAS sampled particulate matter onto a 37-mm-diameter polytetrafluoroethylene (PTFE) membrane filter (PT37P-PF03, Measurement Technology Laboratories, Minneapolis, MN, USA) at 2 L min^-1, through either a PM_2.5 or PM₁₀ inlet, for one week at a time.  Samples were started on September 3^rd and September 10^th.  All UPAS samples started and ended at the same time each week.  Each filter was weighed in triplicate, to the nearest 1 μg, before and after the sample on a balance (XS3DU, Mettler-Toledo, Columbus, OH, USA) in a climate-controlled laboratory facility at Colorado State University. 

PM_2.5 and PM₁₀ concentrations measured by the regulatory monitors at the CAMP site were downloaded using the U.S. EPA Air Quality System (AQS) Application Programming Interface (API).  The 24-h average PM_2.5 concentrations derived from FRM filter samples, 1-h average PM_2.5 concentrations measured by the GRIMM EDM 180, and 1-h average PM₁₀ concentrations measured by the GRIMM EDM 180 were averaged over the duration of each week-long UPAS filter sample, with 24-h and 1-h measurements that only partially overlapped with each UPAS sample weighted accordingly.  Additionally, the 24-h average PM_2.5 concentration derived from the September 13^th FRM filter sample was not available and was replaced with the 24-h average PM_2.5 concentration measured by the GRIMM EDM 180 when calculating the week-long average for September 10^th-17^th.

One-week average PM_2.5 and PM₁₀ concentrations measured using UPAS filter samples are compared to 1-week average PM_2.5 and PM₁₀ concentrations measured by the FRM and FEM instruments in the table below.  Note that ambient PM concentrations in Denver were elevated during this two-week period due to smoke from nearby wildfires.

During the week of September 3^rd-10^th, the mean 1-week-average PM_2.5 concentration calculated from nine replicate UPAS samples was 12.9 μg m^-3.  For comparison, the 1-week PM_2.5 concentration calculated by averaging the 24-h FRM filter samples collected that week was 13.3 μg m^-3.  The mean difference between the 1-week PM_2.5 concentration measured using each UPAS and the 1-week PM_2.5 concentration calculated from the FRM filter samples was -0.4 μg m^-3 or -3.3%.  The relative standard deviation (RSD) of the nine replicate UPAS measurements was 7.7%.  Overall, 1-week-average PM_2.5 concentrations measured by the UPAS and the FRM filter sampler showed excellent agreement.  This result was expected because the UPAS and FRM sampler measured PM_2.5 using the same physical approach: sampling through a PM_2.5 cyclone onto a PTFE filter followed by gravimetric analysis of the sample.

Mean 1-week-average PM_2.5 concentrations measured using UPAS filter samples were slightly lower than 1-week average PM_2.5 concentrations measured by the GRIMM EDM 180.  The mean difference between the 1-week PM_2.5 concentration measured using each UPAS and the 1-week PM_2.5 concentration measured by the GRIMM EDM 180 was -1.9 μg m^-3 (or -12.7%) during the first week and -2.5 μg m^-3 (or -17.5%) during the second week.  We speculate that the worse agreement between the UPAS and the GRIMM EDM 180 was due to the different physical methods these two instruments used to measure PM_2.5.  The GRIMM EDM 180 is an optical particle counter and relies on assumptions about particle density to calculate PM mass concentrations from particle size and count data.  If a large portion of the PM_2.5 present during these two weeks was wildfire smoke that had a higher density than typical urban PM (see Table 2 in Hagan and Kroll [2020]), the GRIMM EDM 180 might have overestimated  concentrations slightly.

The 1-week-average PM₁₀ concentrations measured using UPAS filter samples were similar to those measured by the GRIMM EDM 180. Excellent agreement between the two sets of measurements was observed during the week of September 10^th-17^th, when the mean 1-week-average PM₁₀ concentration calculated from ten replicate UPAS samples was 24.8 μg m^-3 and the 1-week-average PM₁₀ concentration calculated from 1-h-average GRIMM measurements was 25.0 μg m^-3.  The mean difference between the 1-week PM₁₀ concentration measured using each UPAS and the 1-week PM₁₀ concentration calculated from the GRIMM EDM 180 data was -0.1 μg m^-3 or -0.5%. The relative standard deviation (RSD) of the ten replicate UPAS PM₁₀ measurements was 3.4%.

Further reading

In our public Zotero library, you can find peer-reviewed publications in which particulate matter concentrations measured with UPAS were compared to concentrations measured using more conventional air samplers.  Several of these publications focused specifically on stationary ambient monitoring. For example, Rodriguez-Villamizar et al. (2023) sampled PM_2.5 using UPAS v2.0 that were equipped with 1 L min^-1 inlets and collocated with beta attenuation monitors (BAMs) within local air quality networks in Colombian cities.  The agreement between 1-week average PM_2.5 concentrations measured using UPAS and BAMs is discussed on page 3212 of that publication and illustrated in Figure S1.