Comparison of Different Hybrid Modeling Methods for the Estimation of Intraurban Air Pollutant Concentrations

Abstract

Exposure to air pollution – a major potential environmental factor – can exert significant health effects on the urban population. The improvement of modeling methods to estimate the concentrations of air pollutants within a complex city should be important for exposure assessment of subjects in health studies. This paper presents several hybrid modeling approaches to simulate highly resolved variability in ambient air pollution in the capital of South Korea, Seoul. They combine the Community Multiscale Quality (CMAQ), a regional photochemical model with fine scale models including the California Puff (CALPUFF) dispersion model and the land use regression (LUR) model. We compared the high-resolution spatial pollutant concentration estimates from four different hybrid combinations: 1) raw CMAQ-CALPUFF; 2) observation-fused CMAQ-CALPUFF; 3) raw CMAQ-LUR; 4) observation-fused CMAQ-LUR. We quantitively evaluated the simulated concentrations with field data from mobile measurements carried out during the winter season. The results showed that significant differences in sub-grid variability of pollutant concentrations were found according to different hybrid modeling methods and observation-fused hybrid modeling can generally improve the model performance in a complex urban area. Our study suggests that a properly evaluated hybrid modeling approach could increase the accuracy of air pollutant concentration estimates for the purpose of improving exposure assessment in a health study.