Combined assimilation of NOAA surface and MIPAS satellite observations to constrain the global budget of carbonyl sulfide

Carbonyl sulfide (inline-formulaCOS), a trace gas in our atmosphere that leads to the formation of aerosols in the stratosphere, is largely taken up by terrestrial ecosystems. Quantifying the biosphere uptake of inline-formulaCOS could provide a useful quantity to estimate gross primary productivity (GPP). Some inline-formulaCOS sources and sinks still contain large uncertainties, and several top-down estimates of the inline-formulaCOS budget point to an underestimation of sources, especially in the tropics. We extended the inverse model TM5-4DVAR to assimilate Michelson Interferometer for Passive Atmospheric Sounding (MIPAS) satellite data, in addition to National Oceanic and Atmospheric Administration (NOAA) surface data as used in a previous study. To resolve possible discrepancies among the two observational data sets, a bias correction scheme is necessary and implemented. A set of inversions is presented that explores the influence of the different measurement streams and the settings of the prior fluxes. To evaluate the performance of the inverse system, the HIAPER Pole-to-Pole Observations (HIPPO) aircraft observations and NOAA airborne profiles are used. All inversions reduce the inline-formulaCOS biosphere uptake from a prior value of 1053 inline-formulaGgS ainline-formula⁻¹ to much smaller values, depending on the inversion settings. These large adjustments of the biosphere uptake often turn parts of Amazonia into a inline-formulaCOS source. Only inversions that exclusively use MIPAS observations, or strongly reduce the prior errors on the biosphere flux, maintain the Amazon as a inline-formulaCOS sink. Inclusion of MIPAS data in the inversion leads to a better separation of land and ocean fluxes. Over the Amazon, these inversions reduce the biosphere uptake from roughly 300 to 100 inline-formulaGgS ainline-formula⁻¹, indicating a strongly overestimated prior uptake in this region. Although a recent study also reported reduced inline-formulaCOS uptake over the Amazon, we emphasise that a careful construction of prior fluxes and their associated errors remains important. For instance, an inversion that gives large freedom to adjust the anthropogenic and ocean fluxes of inline-formulaCS₂, an important inline-formulaCOS precursor, also closes the budget satisfactorily with much smaller adjustments to the biosphere. We achieved better characterisation of biosphere prior and uncertainty, better characterisation of combined ocean and land fluxes, and better constraint of both by combining surface and satellite observations. We recommend more inline-formulaCOS observations to characterise biosphere and ocean fluxes, especially over the data-poor tropics.