Articles | Volume 13, issue 21
Atmos. Chem. Phys., 13, 11005–11018, 2013
Atmos. Chem. Phys., 13, 11005–11018, 2013

Research article 12 Nov 2013

Research article | 12 Nov 2013

Inverse modeling of Texas NOx emissions using space-based and ground-based NO2 observations

W. Tang1, D. S. Cohan1, L. N. Lamsal3,2, X. Xiao1, and W. Zhou1 W. Tang et al.
  • 1Department of Civil and Environmental Engineering, Rice University, 6100 Main Street MS 519, Houston, TX 77005, USA
  • 2NASA Goddard Space Flight Center, Greenbelt, MD, USA
  • 3Goddard Earth Sciences Technology & Research, Universities Space Research Association, Columbia, MD, USA

Abstract. Inverse modeling of nitrogen oxide (NOx) emissions using satellite-based NO2 observations has become more prevalent in recent years, but has rarely been applied to regulatory modeling at regional scales. In this study, OMI satellite observations of NO2 column densities are used to conduct inverse modeling of NOx emission inventories for two Texas State Implementation Plan (SIP) modeling episodes. Addition of lightning, aircraft, and soil NOx emissions to the regulatory inventory narrowed but did not close the gap between modeled and satellite-observed NO2 over rural regions. Satellite-based top-down emission inventories are created with the regional Comprehensive Air Quality Model with extensions (CAMx) using two techniques: the direct scaling method and discrete Kalman filter (DKF) with decoupled direct method (DDM) sensitivity analysis. The simulations with satellite-inverted inventories are compared to the modeling results using the a priori inventory as well as an inventory created by a ground-level NO2-based DKF inversion. The DKF inversions yield conflicting results: the satellite-based inversion scales up the a priori NOx emissions in most regions by factors of 1.02 to 1.84, leading to 3–55% increase in modeled NO2 column densities and 1–7 ppb increase in ground 8 h ozone concentrations, while the ground-based inversion indicates the a priori NOx emissions should be scaled by factors of 0.34 to 0.57 in each region. However, none of the inversions improve the model performance in simulating aircraft-observed NO2 or ground-level ozone (O3) concentrations.

Final-revised paper