Predicting Near-Road PM 2.5 Concentrations

Scientific evidence has increasingly shown an association between particulate matter (PM) and adverse human health impacts. Accurately predicting near-road PM_2.5 concentrations is therefore important for project-level transportation conformity and health risk analysis. This study assessed the capability and performance of three dispersion models–-CALINE4, CAL3QHC, and AERMOD–-in predicting near-road PM_2.5 concentrations. The comparative assessment included identifying differences among the three models in relation to methodology and data requirements. An intersection in Sacramento, California, and a busy road in London were used as sampling sites to evaluate how model predictions differed from observed PM_2.5 concentrations. Screen plots and statistical tests indicated that, at the Sacramento site, CALINE4 and CAL3QHC performed moderately well, while AERMOD under-predicted PM_2.5 concentrations. For the London site, both CALINE4 and CAL3QHC resulted in overpredictions when incremental concentrations due to on-road emission sources were low, while underpredictions occurred when incremental concentrations were high. The street canyon effect and receptor location likely contributed to the relatively poor performance of the models at the London site.