A land use regression model for predicting ambient fine particulate concentrations across Los Angeles.

摘要

This study uses geographic information systems (GIS) to integrate data from land use, transportation and physical geography to derive a fine particulate (PM2.5) pollution surface for Los Angeles. The EPA had 23 monitors of PM2.5 for the study year 2000. Multivariate linear regression was used to create base models for the PM2.5 surface. Regression models were applied to 18000 lattice points. Inverse distance weighting was used to produce a spatially continuous surface. The best model explained 83% of the variance in PM2.5 using industrial, government and commercial areas, and collector and arterial roads as predictors, showing elevated concentrations of PM2.5 in both the central city and surrounding the ports in Long Beach. Exposure estimates will be utilized in future research to test the relation between atherosclerosis and air pollution.