Gauging intraurban variability of ambient particulate matter arsenic and other air toxic metals from a network of monitoring sites

Abstract

A four site monitoring network was established in the Missouri portion of Metropolitan St. Louis during 2008 to characterize spatiotemporal patterns in PM10 arsenic. Arsenic measured at two urban sites in the City of St. Louis was typically higher than arsenic at two suburban sites. Spatiotemporal variability in arsenic is examined by plotting the Pearson correlation coefficient (PCC) against the coefficient of divergence (COD) for each site-pair to merge the temporal tracking ability of PCC with COD's ability to gauge spatial homogeneity. Arsenic measured across the network is apportioned into a network-wide baseline and site-specific excess concentrations to semi-quantitatively differentiate local-scale emission source contributions from sources exerting influence over larger spatial scales. Comparing measured concentrations at each site against the network-wide baseline concentration using a scattergram of PCC and COD emphasizes the impact of local sources on intraurban variability. Conditional probability function (CPF) plots constructed using site-specific measured arsenic and surface winds identify a broad emission source region towards the east, but mask the bearings of local sources in the urban core. CPF plots using site-specific arsenic in excess of the baseline concentrations provide better resolution of local emission source bearings and are triangulated to identify a likely arsenic emission source zone along the industrialized Mississippi Riverfront. Additional air toxic metals measured in this study (selenium, manganese and lead) are also investigated to examine the efficacy of this methodology to characterize intraurban variability.