Intraurban Spatiotemporal Variability of Ambient Air Pollutants across Metropolitan St. Louis.

摘要

Ambient air monitoring networks have been established in the United States since the 1970s to comply with the Clean Air Act. The monitoring networks are primarily used to determine compliance but also provide substantive support to air quality management and air quality research including studies on health effects of air pollutants. The Roxana Air Quality Study (RAQS) was conducted at the fenceline of a petroleum refinery in Roxana, Illinois. In addition to providing insights into air pollutant impacts from the refinery, these measurements increased the St. Louis area monitoring network density for gaseous air toxics and fine particulate matter (PM2.5) speciation and thus provided an opportunity to examine intraurban spatiotemporal variability for these air quality parameters.;This dissertation focused on exploring and assessing aspects of ambient air pollutant spatiotemporal variability in the St. Louis area from three progressively expanded spatial scales using a suite of methods and metrics. RAQS data were used to characterize air quality conditions in the immediate vicinity of the petroleum refinery. For example, PM2.5 lanthanoids were used to track impacts from refinery fluidized bed catalytic cracker emissions. RAQS air toxics data were interpreted by comparing to network data from the Blair Street station in the City of St. Louis which is a National Air Toxics Trends Station. Species were classified as being spatially homogeneous (similar between sites) or heterogeneous (different between sites) and in the latter case these differences were interpreted using surface winds data. For PM 2.5 species, there were five concurrently operating sites in the St. Louis area - including the site in Roxana - which are either formally part of the national Chemical Speciation Network (CSN) or rigorously follow the CSN sampling and analytical protocols. This unusually large number of speciation sites for a region the size of St. Louis motivated a detailed examination of these data. Intraurban spatiotemporal variability for certain species was evaluated in the context of measurement error. For example, for species otherwise considered homogeneous, differential impacts from local point sources at different locations could be identified after comparing the observed day-to-day variations to those contributed by measurement error. In addition, source apportionment modeling was conducted using single- and multi-site datasets to assign measured PM2.5 mass to emission source categories. A suite of approaches were used to aid in the selection of an appropriate number of factors including metrics recently added to the US EPA Positive Matrix Factorization (EPA PMF) modeling software and the sensitivity of modeling results to perturbations on the measurement uncertainties.