The study intended to develop a land use regression model to predict inhalable particulate matter (PM10) concentrations in São Paulo, Brazil. The model was based on the 2007 average of PM10 available from 9 monitoring stations and demographic, roads and land use data in buffers of 250 to 1,000 m. Simple linear regression were used to select the more robust independent variables and those without collinearity. Four variables were selected for the multiple regression analysis. Only light traffic in buffers <250 m remained in the final model which explained 63.8% of the PM10variance.The land use regression approach is a quick and easy method to predict air pollution levels. However, our model was based on measurements of only a few sites.
air pollution; environmental exposure; particulate matter; environmental impact; environmental health
