PM2.5 Exposure Mapping in the U.S. Integrating Land Use Regression Models

• Activity Number: 16
• Type: Topic Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Seung-Jae Lee*+ and Bernie Beckerman and Micheal Jerrett
• Companies: University of California at Berkeley and University of California at Berkeley
• Keywords: Air Pollution ; Land Use Regression ; Bayesian Maximum Entropy ; Exposure Mapping

Abstract

We develop a methodology for representing air pollutant levels at fine spatiotemporal scales and validate it through the example of the PM2.5 distribution in the U.S. This work takes advantage of refined Land Use Regression (LUR) models embodying the small area variation of PM2.5 dominated by near roadways and industrial facilities. The LUR models show improved capabilities to identify mean trend characteristics of PM2.5, processed through Bayesian Maximum Entropy for the estimation of monthly PM2.5 for the entire U.S. The LUR-contained estimation improves upon LUR-free estimation as well as remote sensing estimates of PM2.5. It provides, therefore, crucial space/time characteristics for improved decision making on the relationship between air pollution exposure and adverse health outcomes.