Exposure Assessment of Ambient NO2 Using Satellite Ozone Monitoring Instrument (OMI) NO2 and Land Use Regression

Abstract

Exposure Assessment of Ambient NO2 Using Satellite Ozone Monitoring Instrument (OMI) NO2 and Land Use RegressionAbstract Number:2690 Hyung Joo Lee* and Petros Koutrakis Hyung Joo Lee* National Aeronautics and Space Administration Postdoctoral Program, National Aeronautics and Space Administration Ames Research Center, United States, E-mail Address: [email protected] Search for more papers by this author and Petros Koutrakis Harvard School of Public Health, United States, E-mail Address: [email protected] Search for more papers by this author AbstractThe exposure assessment of ambient air pollution is generally performed by using ground measurements from central monitoring sites. However, this conventional approach is likely to cause exposure errors because of sparsely distributed ground monitors. Satellite remote sensing data have been increasingly used to assess subjects’ exposures to ambient air pollutants including but not limited to NO2. In this study, we estimated daily ground-level NO2 concentrations using satellite Ozone Monitoring Instrument (OMI) NO2 data and land use regression in the New England region, U.S., for the period 2005-2010. A mixed effects model was constructed to calibrate satellite tropospheric NO2 densities for ground NO2 concentrations on a daily basis. In addition, fine-scale land use parameters including population density, distance to major highways, elevation, and percent developed area were used to reflect the impact of local sources on ambient NO2 concentrations, which enabled us to estimate NO2 concentrations at point locations. The cross-validation mixed effects model showed the coefficient of determination (R2) of 0.79 and root-mean-squared-error (RMSE) of 3.59 ppb between the measured and estimated NO2 concentrations. Furthermore, NO2 concentration maps using the estimated NO2 concentrations clearly demonstrated high pollution levels in high populated/traffic areas and season-specific pollutant characteristics. This study suggests that the modeling approach combining satellite NO2 with land use regression can lead us to spatially and temporally resolved exposure assessment of ambient NO2 for epidemiological studies.