From ground to orbit: Improving global methane emission inversions with adaptive weighting of TROPOMI and NOAA data

Methane (CH₄) has a relatively short, compared to other greenhouse gases, atmospheric lifetime (~9 years) and a highly effective radiative forcing, contributing ~31% (1.19 W m-2) of the additional radiative forcing from anthropogenic emissions during the industrial era. Therefore, reducing CH₄ emissions is a necessary target for limiting near-term climate change. Despite progress in understanding processes and improving estimates, uncertainties in CH₄ sources and sinks create discrepancies between bottom-up and top-down estimates. Recent satellites carrying high-resolution, accurate instruments have provided better information on the concentrations and distributions of atmospheric trace gases. High-confidence observations and in-situ measurement networks are essential to reduce discrepancies and increase confidence in the results of data-driven methods.

This study evaluates the feasibility and efficiency of performing inversions integrating satellite-based remote observations, weighted according to their temporal and spatial distribution with in-situ measurements. The inversions assimilate retrieved data from the high-resolution TROPOMI WFMD methane product and background stations from the NOAA network. Compared to the operational ESA product, the TROPOMI WFMD product provides enhanced global daily coverage, especially at higher latitudes, and more realistic uncertainty estimates, offering better insight into the distribution of methane concentrations. Results compare different inversions of global methane emissions for 2019 at 1° × 1° resolution, with and without an adaptive per-pixel weighting factor, performed utilising the TMVar (TM5-MP/4DVar) system. The adaptive inflation factor is applied to the satellite term of the cost function, balancing its contribution relative to the smaller data volume of station measurements. Data from the ground-based TCCON network validate the simulations. This network operates as a standard reference for atmospheric chemistry-transport models and satellite retrievals due to its precision and low uncertainty in total column values.

Preliminary results show no significant differences when including satellite data with the in-situ ones without applying a weighting factor, mainly due to the overwhelming volume of satellite data compared to station measurements. However, introducing a constant weighting factor for satellite observations improves inversion accuracy. Adding an adaptive weighting factor, adjusted based on the observations' temporal and spatial distribution, further enhances the results. This approach outperforms unweighted and constant-weighting methods by addressing the underuse of station data and neglect of satellite information in regions with lower coverage due to lower point density. Therefore, incorporating appropriately weighted sources of information on the total atmospheric state helps to optimise the inversion results and ultimately reduce the error in the constrained surface fluxes.