Incorporation of Satellite Observations into Texas Ozone Attainment Modeling
Cohan, Daniel S.
Doctor of Philosophy
Uncertain photolysis rates and nitrogen oxides (NOx) emission inventories impair the accuracy of ozone (O3) regulatory modeling. Satellite-observed clouds have been used to correct model-predicted photolysis rates, and satellite-constrained top-down NOx emissions have been used to identify and reduce uncertainties in bottom-up NOx emissions. However, studies on using multiple satellite-derived model inputs to improve O3 State Implementation Plan (SIP) modeling are rare. In this thesis, observations of clouds from the Geostationary Operational Environmental Satellite (GOES) and of NO2 from the Ozone Monitoring Instrument (OMI) are used to adjust the inputs to SIP modeling of O3 in Texas. The discrete Kalman filter (DKF) inversion approach is successfully applied with decoupled direct method (DDM) sensitivities in the Comprehensive Air Quality Model with extensions (CAMx) model to adjust Texas NOx emissions in designated emission regions and categories to better match OMI NO2 data. The NO2 vertical column densities (VCD) gap between OMI and CAMx over rural areas is alleviated by adding missing lightning and aviation and underestimated soil NOx emissions to the base regulatory emission inventory and further reduced by increasing modeled NOx lifetime and adding an artificial NO2 layer in the upper troposphere. The region-based DKF inversion using OMI NO2 tends to scale up NOx emissions in most regions, which conflicts with the inversion results using ground NO2 measurements and fails to improve the ground-level O3 simulations. The sector-based DKF inversion using OMI NO2 suggests scaling down area and non-road NOx emissions by 50%, leading to approximately 2-5ppb decrease in ground 8-h O3 concentrations, and improving both hourly ground-level NO2 and O3 simulations by reducing biases by 0.25 and 0.04 and errors by 0.13 and 0.04, respectively. Finally, using both GOES-derived photolysis rates and OMI-constrained NOx emissions reduces modeled bias and error by 0.05, and increases the model correlations in simulating ground O3 measurements and makes O3 more sensitive to NOx emissions in the O3 nonattainment areas.