A hybrid approach to estimating national scale spatiotemporal variability of PM2.5 in the contiguous United States

Environ Sci Technol. 2013 Jul 2;47(13):7233-41. doi: 10.1021/es400039u. Epub 2013 Jun 11.


Airborne fine particulate matter exhibits spatiotemporal variability at multiple scales, which presents challenges to estimating exposures for health effects assessment. Here we created a model to predict ambient particulate matter less than 2.5 μm in aerodynamic diameter (PM2.5) across the contiguous United States to be applied to health effects modeling. We developed a hybrid approach combining a land use regression model (LUR) selected with a machine learning method, and Bayesian Maximum Entropy (BME) interpolation of the LUR space-time residuals. The PM2.5 data set included 104,172 monthly observations at 1464 monitoring locations with approximately 10% of locations reserved for cross-validation. LUR models were based on remote sensing estimates of PM2.5, land use and traffic indicators. Normalized cross-validated R(2) values for LUR were 0.63 and 0.11 with and without remote sensing, respectively, suggesting remote sensing is a strong predictor of ground-level concentrations. In the models including the BME interpolation of the residuals, cross-validated R(2) were 0.79 for both configurations; the model without remotely sensed data described more fine-scale variation than the model including remote sensing. Our results suggest that our modeling framework can predict ground-level concentrations of PM2.5 at multiple scales over the contiguous U.S.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Air Pollutants / analysis*
  • Bayes Theorem
  • Environmental Monitoring / statistics & numerical data*
  • Models, Theoretical*
  • Motor Vehicles
  • Particle Size
  • Particulate Matter / analysis*
  • Regression Analysis
  • Remote Sensing Technology
  • Spatio-Temporal Analysis
  • United States


  • Air Pollutants
  • Particulate Matter