An optimized spatial proximity model for fine particulate matter air pollution exposure assessment in areas of sparse monitoring

Building upon the utility of spatial proximity models and the theoretical reliability of Gaussian dispersion processes of air pollutants, this study puts forward a novel Gaussian weighting function-aided proximity model GWFPM. The study area and data set for this work consisted of transport-related emission sources of PM2.5 in the Houston-Baytown-Sugar Land metropolitan area. Performance of the GWFPM was validated by comparing on-site observed PM2.5 concentrations with results from classical ordinary kriging OK interpolation and a robust emission-weighted proximity model EWPM. Results show that the fitting R2 between possible exposure intensity calculated by GWFPM and observed PM2.5 concentrations was 0.67. A variety of statistical evidence i.e., bias, root mean square error [RMSE], mean absolute error [MAE], and correlation coefficient confirmed that GWFPM outperformed OK and EWPM in estimating annual PM2.5 concentrations for all monitoring sites. These results indicate that a GWFPM utilizing the physical dispersing mechanisms integrated may more effectively characterize annual-scale exposure than traditional models. Using GWFPM, receptors’ exposure to air pollution can be assessed with sufficient accuracy, even in those areas with a low density of monitoring sites. These results may be of use to public health and planning officials in a more accurate assessment of the annual exposure risk to a population, especially in areas where monitoring sites are sparse.