Modeling Exposure to Traffic-Related Air Pollutants for the Residential Human Health Risk Assessment Study in Kyiv, Ukraine

摘要

Academy of Medical Sciences of Ukraine", Kyiv, Ukraine Arina Petrosian, State Institution "O.M. Marzeyev Institute for Public Health of the National Academy of Medical Sciences of Ukraine", Kyiv, Ukraine Introduction: There is a strong body of evidence linking traffic-related air pollution exposure to adverse health effects. Although accurate air quality data is rarely available for the urban residential neighborhood. This provides the ground for the development of various modeling techniques for obtaining more accurate exposure estimates compared to monitoring data. Assessment of human health risks attributable to traffic related air pollution study was designed to evaluate the risks of development of respiratory health outcomes in population living near roadways in Kyiv, Ukraine. Methods: A combination of modeling approaches was used to estimate exposures to traffic-related air pollutants for the residents living in close proximity to the roadways in two city districts. Field studies were conducted to provide measurement data for developing, evaluating and refining the models. CO, NO2, SO2, NMHC, PM2,5, PM10, PM1 concentrations, meteorological parameters, fleet intensity and structure data were obtained in two measurement campaigns (2012-2014). A land-use regression (LUR) models were developed to capture NO2, CO, PM10, PM2,5, PM1 near-road variability. I5C-AERMOD View air pollution dispersion model was used to model averaged 1-, 8-, 24-hour and annual concentrations of CO, NO2, SO2, NMHC and PM10 at 952 receptor points covering the area of the study. Agreement between ISC-AERMOD View modeled and measured concentrations was assessed based on the standard methods. Results: Vehicles number, road distance, road length, land use and meteorological variables were the most important predictors of traffic-related air pollution variability for LUR models. Developed models explained up to 75% of the spatial variability for studied pollutants in near-roadway residential neighborhoods and demonstrated a good agreement between observed concentrations and predicted levels (average difference of 3-10%). Correlation analysis revealed strong association between ISC-AERMOD view modeled values and measurements of NO2, (r=0,93, p<0,005) and weaker associations for other pollutants (p<0,005): r=0,88 for CO, r=0,90 for SO2, r=0,88 for NMHC, and r=0,74 for PM10. Conclusions: The results demonstrated an integrated measurement study design. Applied modeling techniques proved to be an effective instrument of human exposure assessment when limited monitoring data is available. Based on the models outputs traffic-related exposure maps were developed for the study area and corresponding human health risks formed by traffic- related air pollution were assessed.