Mitigating the Impact of Chemical Transport Model Biases on Top-down CO and NOx Emission Estimates using Multi-species Chemical Data Assimilation

Chemical transport models such as GEOS-Chem simulate tropospheric CO-HOx-NOx-O3 chemistry, which determines the oxidizing power of the atmosphere as well as the abundance of tropospheric CO, NO2, and O3. Inverse modelling analyses to estimate CO and NOx emissions traditionally rely only on observations of the species of interest, neglecting chemical biases in the models employed in the analyses. However, top-down estimates of CO and NOx emissions are sensitive to tropospheric OH and O3, respectively. In this thesis, two approaches were used to mitigate the model biases to estimate CO and NOx emission. First, we developed a multi-species data assimilation framework using the GEOS-Chem 4D-Var data assimilation system. Satellite observations of CO, NO2, O3, HCHO, and HNO3 were assimilated to optimize CO and NOx emissions, as well as O3, HCHO and HNO3 initial conditions. After the assimilation, improved chemical states were produced with CO, NO2, and O3 fields that were in closer agreement with independent observations. The multi-species DA also provided a means of indirectly constraining the tropospheric OH. The resulting global mean tropospheric OH was more consistent with independent empirically derived OH estimates. We also examined the a posteriori CO and NOx emissions. During the study period, my thesis showed the a priori CO emissions were underestimated and the a priori NOx emissions were overestimated in the northern extratropics. The emission estimates in Europe, the Amazon, and equatorial Africa were most sensitive to the multi-species information assimilated into GEOS-Chem. Second, we assessed the utility of a new OMI-minus-OSIRIS (OmO) NO2 retrieval product for improving North American NOx emission estimates. We assimilated the OmO data using the nested version of GEOS-Chem with assimilated O3 boundary conditions. The results suggested that both the anthropogenic and biomass burning sources were significantly overestimated in the a priori. The changes in the a posteriori NOx emissions inferred from the OmO data between 2008 and 2011 were found to be in better agreement with the changes in independent NO2 observations than those estimated from OMI data.Ph.D