A Spatial Investigation of Coarse Particulate Mass and Components During Two Seasons in St Paul, MN

TIMOTHY V. LARSON (1), David R. Jacobs (2), Adam A. Szpiro (1), Amanda Gassett (1), Joel D. Kaufman (1), Sara D. Adar (1)

(1) University of Washington, Seattle, (2) University of Minnesota, Minneapolis

     Abstract Number: 387
     Last modified: November 9, 2009

Abstract
Two-week average PM$_(10) and PM$_(2.5) co-located filter samples were collected over two different seasons at approximately 20 locations in St. Paul, MN as part of a larger EPA-funded study on the health effects of chronic exposure to coarse PM. Coarse mass concentrations were computed by difference and had a mean value of 3.9 micrograms/meter$^3 (range below detection to 11.5 micrograms/meter$^3) in winter and 7.6 micrograms/meter$^3 (range 4.4 to 15.7 micrograms/meter$^3) in summer. During the winter sampling period, the ground was covered with snow. A simple land use regression (LUR) model was employed using a number of spatial covariates to predict the variation of the PM$_(10-2.5) across the urban area. Summer concentrations were only associated with a few traffic variables, including the sum of A1 road lengths (CFCC classification system) within 500 meters and A2 road lengths within 100 meters (R$^2 = 0.75, RMSE = 1.6 micrograms/meter$^3). Winter concentrations were associated with the sum of A1 road lengths within 750 meters and residential population within 150 meters (R$^2 = 0.45, RMSE = 2.0 micrograms/meter$^3). Positive matrix factorization was performed on the winter samples using a standard suite of trace elements via XRF. A four factor model was obtained with most of the coarse mass associated with a single factor with high relative loadings of Na, Cl, Fe, Ni, Cu, and Ba and a distinct absence of Ca and Si. We interpret this as a winter road salt factor. The LUR model for this feature had slightly improved performance over coarse mass alone (R$^2 = 0.53) and included the sum of A1 and A3 road lengths within 750 meters.