Using a network of lower-cost monitors to identify the influence of modifiable factors driving spatial patterns in fine particulate matter concentrations in an urban environment

Background There is substantial interest in using networks of lower-cost air quality sensors to characterize urban population exposure to fine particulate matter mass (PM 2.5 ). However, sensor uncertainty is a concern with these monitors. Objectives (1) Quantify the uncertainty of lower-cost PM 2.5 sensors; (2) Use the high spatiotemporal resolution of a lower-cost sensor network to quantify the contribution of different modifiable and non-modifiable factors to urban PM 2.5 . Methods A network of 64 lower-cost monitors was deployed across Pittsburgh, PA, USA. Measurement and sampling uncertainties were quantified by comparison to local reference monitors. Data were sorted by land-use characteristics, time of day, and wind direction. Results Careful calibration, temporal averaging, and reference site corrections reduced sensor uncertainty to 1 μg/m 3 , ~10% of typical long-term average PM 2.5 concentrations in Pittsburgh. Episodic and long-term enhancements to urban PM 2.5 due to a nearby large metallurgical coke manufacturing facility were 1.6 ± 0.36 μg/m 3 and 0.3 ± 0.2 μg/m 3 , respectively. Daytime land-use regression models identified restaurants as an important local contributor to urban PM 2.5 . PM 2.5 above EPA and WHO daily health standards was observed at several sites across the city. Significance With proper management, a large network of lower-cost sensors can identify statistically significant trends and factors in urban exposure.