Comparison of Low-Cost PM Sensors with a Direct Mass Measurement at Sites Representing Road and Non-road PM Sources
YU JUNG LIN, Karl Armstrong, Roby Greenwald,
Georgia State University Abstract Number: 480
Working Group: Health-Related Aerosols
AbstractThe use of low-cost PM microsensors for time-resolved personal, occupational, or residential exposure is rapidly increasing. Given that these sensors are typically based on optical methods of detection, inferring mass concentration requires assumptions of particle shape, density, refractive index, and hygroscopic growth. These factors are related to source-specific particle composition as well as meteorological factors including temperature (T) and relative humidity (RH). On the other hand, the Tapered Element Oscillating Microbalance (TEOM) directly measures PM mass with high time-resolution and controlled inlet environmental conditions.
We deployed multiple units of three common microsensor models at TEOM-equipped monitoring stations operated by the Georgia Environmental Protection Division. Each sample site is located in a different urban context representing a range of sources: directly adjacent to a 16-lane urban freeway, intermediate-distant (500m) to an 8-lane suburban highway, and an urban background site over 3 km from the nearest highway. We collected microsensor, TEOM, and meteorological data over three days at each site under a variety of meteorological conditions in all four seasons and aggregated all data to 1-hour time resolution. We used multivariate regression models to estimate coefficients for site-specific meteorological and microsensor variables with TEOM data as the reference and assessed model performance using a ten-fold cross validation procedure.
As expected, the mass concentration reported by microsensor devices is substantially different than TEOM measurements and is correlated with ambient T and RH. We observed meaningful differences in regression coefficients across sample sites reflecting variation in hygroscopicity for particles from different sources. Additionally, there is evidence of a seasonal influence on regression coefficients, perhaps a result of the differing levels of biogenic or secondary organic aerosols in this region.