Comparison of NOx Fluxes Measured by Eddy Covariance to Emission Inventories and Land Use

Uncertainty in emission inventories remains a critical limitation of air quality modeling and management. Using eddy covariance, we measured surface-atmosphere exchange fluxes of nitrogen oxides (NOx) at the neighborhood scale at 13 sites in the Norfolk, Virginia area to estimate emissions, to evaluate official inventories, and to quantify relationships between emissions and land use. Average daytime fluxes ranged from 0.4 mu g m(-2) s(-1) at a site near open water to 9.5 mu g m(-2) s(-1) at a site dominated by vehicle traffic. NOx fluxes were correlated with both road density and medium- plus high-intensity development, confirming that both motor vehicles and sources associated with development are responsible for NOx emissions in urban areas. Spatially averaged NOx fluxes measured by eddy covariance agreed to within 3% with the National Emission Inventory (NEI) but were 2.8 times higher than those in the corresponding grid cell of an emission inventory developed for air quality modeling. These average fluxes were 4.6, 4.5, and 1.7 mu g m(-2) s(-1), respectively. Uncertainty in the inventories appears to be dominated by the nonroad mobile source category. It is especially important to know NOx emissions accurately because in certain photochemical regimes, reducing NOx emissions can exacerbate secondary pollutant formation.