Large vertical gradient of reactive nitrogen oxides in the boundary layer: Modeling analysis of DISCOVER-AQ 2011 observations

An often used assumption in air pollution studies is a well-mixed boundary layer (BL), where pollutants are evenly distributed. Because of the difficulty in obtaining vertically resolved measurements, the validity of the assumption has not been thoroughly evaluated. In this study, we use more than 200 vertical profiles observed in the Deriving Information on Surface Conditions from Column and Vertically Resolved Observations Relevant to Air Quality (DISCOVER-AQ) aircraft campaign in July 2011 to examine the vertical distributions of pollutants over the Washington-Baltimore area. While many long-lived species are well mixed in daytime, the observed average vertical profile of NOx shows a large negative gradient with increasing altitude in the BL. Our analysis suggests that the magnitude of the NOx gradient is highly sensitive to atmospheric stability. We investigate how parameterizations of the BL and land-surface processes impact vertical profiles in a 1-D chemical transport model, using three BL schemes (Asymmetric Convective Model version 2 (ACM2), Yonsei University (YSU), and Mellor-Yamada-Janjic (MYJ)) and two land-surface schemes (Noah and Rapid Update Cycle (RUC)). The model reasonably reproduces the median vertical profiles of NOx under different BL stability conditions within 30% of observations, classified based on potential temperature gradient and BL height. Comparisons with NOx observations for individual vertical profiles reveal that while YSU performs better in the turbulent and deep BL case, in general, ACM2 (RMSE = 2.0 ppbv) outperforms YSU (RMSE = 2.5 ppbv) and MYJ (RMSE = 2.2 ppbv). Results also indicate that the land-surface schemes in the Weather Research and Forecasting (WRF) model have a small impact on the NOx gradient. Using model simulations, we analyze the impact of BL NOx gradient on the calculation of the ozone production rate and satellite NO2 retrieval. We show that using surface measurements and the well-mixed BL assumption causes a ~45% high bias in the estimated BL ozone production rate and that the variability of NO2 vertical profiles is responsible for 5–10% variability in the retrieved NO2 tropospheric vertical columns.