Long-Term (2005-2017) Global SO2 Emissions Derived from Two OMI SO2 Retrievals: Discrepancy Analysis and Validations

Sulfur dioxide (SO₂) is a reactive air pollutant that leads to acid rain and sulfate aerosol, the latter which contributes to haze and human health problems upon exposure. Bottom-up SO₂ emissions have large uncertainties, and there are often significant lags in time for their availability. Alternatively, SO₂ column densities retrieved from OMI observations provide important information on emission trends and missing sources. However, existing studies show that top-down SO₂ emissions derived from two different OMI SO₂ retrieval products (from NASA and BIRA) are different by a factor of two in China. Here we evaluate these two retrievals as well as a prototype NASA product and attribute the cause of their differences in SO₂ magnitude to usage of different cloud products, which result in different scattering weights and air mass factors in the SO₂ column calculations. We further apply a hybrid 4D-Var / mass balance emissions inversion using the GEOS-Chem adjoint model to derive long term top-down SO₂ emissions with these two (NASA standard and BIRA) products. Global SO₂ gridded emissions are derived at 2° x 2.5° for 2005 – 2017, with higher resolution (0.5° x 0.666°) emissions replaced for East Asia and North America during 2005-2012. Compared to HTAPv2 bottom-up anthropogenic emissions, large changes in SO₂ emissions mainly occur in China, India, Middle East and West US. SO₂ emissions generally decrease since 2007 in China and from 2008 in the US. Top-down emissions are evaluated by comparing SO₂ surface measurements with GEOS-Chem simulated concentrations using posterior emissions from the two products – comparing between the posteriors, simulations with BIRA posterior have smaller bias and better correlation in China and India, whereas NASA posteriors have smaller bias in the US. These analyses explain discrepancies in previous top-down SO₂ emissions and provide new gridded SO₂ emission datasets that better match both satellite observations and surface measurements.