Monitoring and Forecasting Air Quality in the Eastern United States

• Activity Number: 564
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:45 PM
• Sponsor: International Indian Statistical Association
• Author(s): Sujit Sahu*+
• Companies: University of Southampton
• Keywords: Hierarchical Bayesian Modeling ; Spatio-temporal modeling ; Large data sets ; Ozone concentration modeling

Abstract

Estimating exposure to ambient air pollution is an important problem since air pollution data are often collected from several networks of sparsely and irregularly spaced monitoring sites. Data obtained from these sparse networks must be processed using spatial and spatio-temporal methods to check compliance with respect to the relevant national ambient air quality standards at an unmonitored site. Forecasting air pollution, both at the monitoring sites and sites far from the monitors, is also a challenging problem due to the complex nature of dependencies that one needs to model for high dimensional spatio-temporal data. This paper proposes several parametric Bayesian models for spatial interpolation and temporal forecasting of several air pollution species in a vast study region in the eastern United States. Models are compared solely by using their predictive abilities and practicalities in implementation. Validation prediction and forecasting maps from some of the selected models are used to illustrate the methods, some of which are currently being considered by the US Environmental Protection Agency for further public use.