Using different TROPOMI XCH4 retrieval products in atmospheric inversions of CH4: a comparison and reconciliation over Europe

Retrievals of column average methane mixing ratio (XCH₄) from satellites can provide a wealth of information on the fluxes of CH₄. The TROPOMI instrument onboard the satellite Sentinel-5P is particularly interesting for flux estimation using atmospheric inversions as it provides global coverage with relatively high-resolution ground pixels of 5.5 × 7.5 km² (at nadir). Furthermore, there are several different retrieval products available for TROPOMI XCH₄, making it possible to compare these and quantify the effect that differences between them have on CH₄ fluxes from atmospheric inversions.

In this study, we have compared XCH₄ from three retrieval products: i) the official ESA product provided by SRON, ii) the Weighting Function Modified Differential (WFMD) product from the University of Bremen, and iii) the blended TROPOMI-GOSAT product (S5P-BLND) from Harvard University, in atmospheric inversions over Europe for the year 2020. In addition, we compare fluxes from each of these with those derived with an inversion using ground-based observations from the ICOS atmospheric network. For flux estimation, we use the Bayesian inversion framework, FLEXINVERT, and the Lagrangian atmospheric transport model, FLEXPART, driven by ECMWF ERA5 meteorological reanalysis data.

We find that the fluxes derived using WFMD and S5P-BLND products agree better with those derived using ICOS data. The difference in performance of the official ESA product versus the WFMD and S5P-BLND products appears to be related to the albedo bias correction used in the ESA product, which may result in an overcorrection giving too high values of XCH₄ in regions of low albedo and too low values over regions of high albedo. Moreover, this study emphasizes the need to compare different retrieval products and, if available, ground-based observations, in atmospheric inversions to identify any biases and address these.