Daily Ambient NO₂ Concentration Predictions Using Satellite Ozone Monitoring Instrument NO₂ Data and Land Use Regression

Although ground measurements have contributed to revealing the association between ambient air pollution and health effects in epidemiological studies, exposure measurement errors are likely to be caused because of the sparse spatial distribution of ground monitors. In this study, we estimate daily ground NO₂ concentrations in the New England region, U.S., for the period 2005–2010 using satellite remote sensing data in combination with land use regression. To estimate ground-level NO₂ concentrations, we constructed a mixed effects model by taking advantage of spatial and temporal variability in satellite Ozone Monitoring Instrument (OMI) tropospheric column NO₂ densities. Using fine-scale land use parameters, we derived NO₂ concentrations at point locations, which can be further used for subject-specific exposure estimates in epidemiological studies. A mixed effects model showed a reasonably high predictive power for daily NO₂ concentrations (cross-validation <i>R</i>² = 0.79). We observed that the model performed similarly in each season, year, and state. The spatial patterns of model estimates reflected emission source areas (such as high populated/traffic areas) in the study region and revealed the seasonal characteristics of NO₂. This study suggests that a combination of satellite remote sensing and land use regression can be useful for both spatially and temporally resolved exposure assessments of NO<sub>2