Key points are not available for this paper at this time.
Source attribution of volatile organic compounds (VOCs) can be challenging in urban areas, which have many point sources. Mobile laboratories using time-of-flight mass spectrometers (TOF-MS) can take measurements throughout areas of concern, resulting in data with high spatial resolution that can be used to more easily identify these sources. However, emissions in heavily polluted areas still undergo significant mixing over short distances, making source attribution of some compounds challenging. Positive matrix factorization (PMF) has been widely used for attributing pollutants to different sources when taking stationary measurements due to its ability to process large amounts of data into generally interpretable results. However, some limitations of PMF can impact its usefulness to mobile data; PMF is a computationally intensive process, requires some user choices in attributing factors to emissions sources, and results can be significantly impacted by chemical transformations after emission. Here, both PMF and a simpler comparative analysis method are evaluated in analyzing measurements taken in the Elyria Swansea neighborhood of Commerce City, CO. This neighborhood is located near an oil refinery, a wastewater treatment plant, local industrial shops, and major highways. PMF failed to differentiate between oil refinery emissions and traffic emissions, and had difficulties recognizing other key sources. A simpler comparative analysis showed that the refinery contributed significantly to VOC concentrations throughout the neighborhood, including air toxics such as benzene. A wastewater treatment plant contributed to methanethiol and dimethyl sulfide. Finally, a small woodshop was identified as a hyperlocal VOC source, and contributed high amounts of some VOCs, such as toluene and other solvents, in its immediate surroundings.
Rutherford et al. (Mon,) studied this question.