A Novel Sector-Based Inversion to Update NOx, so2 and CO Emissions at the Process Level Using Satellite Observations

Satellite observations have provided insights into the magnitude and trend of air pollutant emissions. However, most studies using these observations to estimate emissions barely consider the correlations of co-emitted species and their chemical interactions, therefore leading to possibly unrealistic estimates. For instance, NO_x, SO₂ and CO emissions are correlated within specific sectors, since each of them is the product of species emission factors and activity rates in the bottom-up estimates. To incorporate their correlations and chemical interactions in the constraints, we developed a novel multi-species sector-based (e.g., energy, industry, transportation, etc.) 4D-Var data assimilation framework using the GEOS-Chem adjoint to estimate emissions according to sector activities. We employed NO₂ and SO₂ satellite observations from OMI and CO observations from MOPITT. The simulated NO₂ and SO₂ concentrations that result from our new sector-based inversion are compared with those from a traditional species-based inversion to evaluate this approach. We find that the NMB of posterior simulations from sector-based inversions are 59.8% (SO₂) and 61.4% (NO₂) smaller than the ones from traditional species-based inversion in China in January 2010. Posterior simulations that constrain activities and species emission factors simultaneously show smaller NMSE than species-only and sector-only inversion frameworks when compared with NO₂ and SO₂ surface measurements. The new framework also enables us to track changes of emission structures over years from a top-down perspective. For example, the activity rate of the industry sector shows a growth rate of 1.58% year^-1 from 2005 to 2012 in China (90% confidence interval); emission factors for transportation CO and transportation NO_x show growth rates of 4.45% year^-1and 0.59% year^-1. Lastly, higher posterior activity rates than the prior in residential, transportation and industry sectors during the winter imply incomplete accounting of anthropogenic activities in the bottom-up inventory. Similarly, the higher posterior emission factors than those from the prior for CO emissions in these three sectors suggest potential underestimate of CO emission rates.