Intraurban-scale dispersion modelling of particulate matter concentrations: Applications for exposure estimates in cohort studies

Epidemiological studies relating air pollution to health effects often estimate personal exposure to particulate matter using values from a central ambient monitoring site as a proxy. However, when there is a significant amount of variation in particulate concentrations across an urban area, the use of central sites may result in exposure misclassification that induces error in long-term cohort epidemiological study designs. When spatially dense monitoring data are not available, advanced dispersion models may offer one solution to the problem of accurately characterising intraurban particulate concentrations across an area. This study presents results from an intraurban assessment of The Air Pollution Model (TAPM)-an Integrated Meteorological-Emission (IME) Model. Particles less than 10 mu m in aerodynamic diameter (PM10) were modelled and compared with a dense intraurban monitoring network in Christchurch, New Zealand, a city with high winter levels of particulate air pollution. Despite the area's high intraurban concentration variability, and meteorological and topographical complexity, the model performed satisfactorily overall, with mean observed and modelled concentrations of 42.9 and 43.4 mu g m(-3), respectively, while the mean Index of Agreement (IOA) between individual sites was 0.60 and the mean systematic RMSE was 16.9 mu g m(-3). Most of the systematic error in the model was due to coarse spatial resolution of the local emission inventory and complex meteorology attributed to localised convergence of drainage flows, especially on the western and southern fringes of the urban area. Given further improvements in site-specific estimates within urban areas, IME models such as TAPM may be a viable alternative to central sites for estimating personal exposure in longer-term (monthly or annual) cohort epidemiological studies. (c) 2005 Elsevier Ltd. All rights reserved.