Jonathan Lesesne

Project title: Multi-scale Spatial Determinants of Fine Particulate Matter (PM) Chemical Composition in US Cities

Degree: MPH | Program: Environmental and Occupational Health (EOH) | Project type: Thesis/Dissertation
Completed in: 2010 | Faculty advisor: Sverre Vedal


Associations between long-term particulate matter (PM) exposure and adverse health effects have been determined in several studies. The National Particle Component Toxicity (NPACT) study is utilizing data on PM chemical species to investigate the effect of PM chemical species on cardiovascular disease in the Women's Health Initiative (WHI). Geographic variation in the chemical composition of PM can likely be explained by geographic features such as measures of traffic as well as other characteristics. The aim of the study is to identify small-scale spatial determinants of PM chemical composition within the United States using a land-use regression (LUR) approach. These spatial determinants will be compared with annual concentrations across the US of five PM measures: PM2.5, Elemental Carbon (EC), Organic Carbon (OC), Sulfate (SO4 -2), and Silicon (Si) obtained from the EPA Speciation Trends Network (STN) air monitoring network. Eventually the LUR model will be used to predict PM chemical species concentrations at locations for which there are no monitoring data. An analysis will then be performed to ascertain if the small-scale spatial determinants of PM species concentrations vary across geographic regions of the United States.

Initial steps of this study include exploring scatter plot relationships between 224 geographic variables and the five PM measures and estimating the respective correlations. Based on theses explorations, a restricted subset of geographic variables were selected for the subsequent analyses. This subset is composed of 31 geographic variables including: meters to coastline; meters to rail yards; buffers of various sizes around A1, A2, and A3 roadways; Normalized Difference Vegetation Index (NDVI) measures, and several others which differ with respect to the individual PM measure. Correlations between these selected geographic variables were then inspected. Multiple linear regression models, regressing PM measure concentration on the restricted subsets of geographic variables, were then generated. Final results of the models are not yet available; however, eventual findings from this study will be useful in estimating residential concentrations of PM chemical species for all women in the WHI cohort. Once obtained, these exposure estimates will be used to estimate affects of long-term exposure to PM chemical species on the incidence of cardiovascular disease and mortality.