Estimation of long-term exposure to air pollution levels over a large spatial domain, such as the mainland UK, entails a challenging modelling task since exposure data are often only observed by a network of sparse monitoring sites with variable amounts of missing data. The paper develops and compares several flexible non-stationary hierarchical Bayesian models for the four most harmful air pollutants, nitrogen dioxide and ozone, and PM10 and PM2.5 particulate matter, in England and Wales during the 5-year period 2007–2011. The models make use of observed data from the UK's automatic urban and rural network as well as output of an atmospheric air quality dispersion model developed recently especially for the UK. Land use information, incorporated as a predictor in the model, further enhances the accuracy of the model. Using daily data for all four pollutants over the 5-year period we obtain empirically verified maps which are the most accurate among the competition. Monte Carlo integration methods for spatial aggregation are developed and these enable us to obtain predictions, and their uncertainties, at the level of a given administrative geography. These estimates for local authority areas can readily be used for many purposes such as modelling of aggregated health outcome data and are made publicly available alongside this paper.
CITATION STYLE
Mukhopadhyay, S., & Sahu, S. K. (2018). A Bayesian spatiotemporal model to estimate long-term exposure to outdoor air pollution at coarser administrative geographies in England and Wales. Journal of the Royal Statistical Society. Series A: Statistics in Society, 181(2), 465–486. https://doi.org/10.1111/rssa.12299
Mendeley helps you to discover research relevant for your work.