A sector-based inversion synergizing NO2, SO2, and CO observations to improve sectoral activity rates and anthropogenic CO2 emissions

Satellite remote sensing provides powerful constraints on atmospheric composition and emissions, yet traditional inversions often rely on single species or instruments and offer limited insight into sector-specific contributions, activity rates, and emission factors needed to inform bottom-up inventories. Here, we develop a sector-based 4D-Var inversion framework that exploits the synergy of multi-instrument, multi-constituent satellite observations to improve atmospheric characterization and emission attribution. By jointly assimilating NO₂, SO₂, and CO observations and leveraging their distinct emission ratios, the framework disentangles contributions from major source sectors, including transportation, industry, residential, aviation, shipping, energy production, biomass burning, soil, and lightning. We assimilate TEMPO NO₂, TROPOMI NO₂ and SO₂, and MOPITT CO observations into the GEOS-Chem adjoint model at 0.25° × 0.3125° resolution over North America. Differences between observations and simulations drive the inversion to optimize sectoral activity rates. Our results reveal large adjustments in lightning and soil NO_x, emphasizing the increasing importance of accurately characterizing background NO₂ to improve air quality simulations. The inversion identifies the transportation sector as the primary contributor to emission adjustments, with top-down transportation emissions 30-60% higher than those in the bottom-up EQUATES inventory along coastal regions and in major urban centers, including Los Angeles, San Francisco, Seattle, Portland, Boston, New York City, and Chicago. These results suggest that recent reductions in transportation emissions may be overestimated in the bottom-up inventory. While TROPOMI and TEMPO NO₂ observations provide consistent constraints on anthropogenic emissions, larger discrepancies are found for natural NOₓ sources, underscoring the importance of synergistic observations for characterizing background atmospheric composition. The framework also enables estimation of sectoral CO₂ emissions using air pollutant observations, extending beyond traditional NOₓ-proxy approaches by capturing industrial and residential emission adjustments through combined SO₂ and CO constraints.