Towards robust global CH₄ emission inversions: Insights into the impact of parametric weighting of TROPOMI observations

Satellite observations are critical for global monitoring of trace gases like methane (CH₄), a potent greenhouse gas, but challenges remain in assimilating dense satellite data alongside sparser in situ measurements within inverse modelling systems. Here, we present a parametric regularisation method that computes observation-specific weights based on the spatial and temporal coverage of satellite data, enabling balanced assimilation across densely and sparsely observed regions. This approach is implemented as a preprocessing step, preserving computational efficiency by maintaining a fixed covariance matrix, and is adaptable for use with multiple satellite products in combined inversions.

Applied to global methane inversions using the TM5-MP/4DVAR system at 1° × 1° resolution for 2019 with TROPOMI observations, our method reduces grid cell weight variability by approximately 20% compared to a constant weighting approach. This adaptation effectively increases the influence of observations from regions with sparse satellite coverage, such as high latitudes and oceans, while reducing over-representation from densely sampled areas. The redistributed weights lead to localised but notable changes in optimised methane fluxes, especially in regions like Southeast Asia and South America, but the global posterior budget remains consistent with the latest Global Methane Budget estimates.

Comparison against independent TCCON and NOAA measurements confirms the robustness of the parametric weighting. Overall, the proposed methodology offers a robust, efficient, and easily generalizable framework for assimilating satellite observations, improving constraints on methane emissions globally, and providing a foundation for future multi-product inversions.