Particulate matter modelling techniques for epidemiological studies of open biomass fire smoke exposure: a review

Abstract

Smoke exposure from landscape and coal mine fires can have severe impacts on human health. The ability of health studies to accurately identify potential associations between smoke exposure and health is dependent on the techniques utilised to quantify exposure concentrations for the population at risk. The evolution of spatial modelling techniques capable of better characterising this association has potential to provide more precise health effect estimates. We reviewed the literature to identify and assess the spatial modelling techniques available to estimate smoke PM 2.5 or PM 10 concentrations from open biomass or coal mine fires. Four electronic databases were searched: MEDLINE, EMBASE, Scopus and Web of Science. Studies were included if they utilised any method for modelling the spatial distribution of PM 2.5 or PM 10 concentrations from open biomass or coal mine fires and had applied the modelled PM to health data. Studies based on un-adjusted monitoring data, or which were not in English, were excluded. We identified 28 studies which utilised five spatial modelling techniques to assess exposure from open biomass fires: dispersion models, land use regression, satellite remote sensing, spatial interpolation and blended models. No studies of coal mine fires were identified. We found the most effective models combined multiple techniques to enhance the strengths and mitigate the weaknesses of the underlying individual techniques. “Blended” models have the potential to facilitate research in regions currently under represented in biomass or coal mine fire studies as well as enhancing the power of studies to identify associations with health outcomes.