High-resolution estimates of national methane emissions for all countries of the world using TROPOMI observations

Observational constraints on national scale methane emissions are needed to assist progress towards the goals of the Paris Agreements and the Global Methane Pledge. Here, we use 2023 blended TROPOMI+GOSAT observations of atmospheric methane in multiple analytical inversions to estimate emissions for all countries of the world at up to 25 km resolution. Prior emissions estimates are spatially distributed according to state-of-the-science bottom-up inventories, and country-level prior totals are adjusted by sector to match the emissions most recently reported to the UNFCCC. We enhance each inversion’s ability to capture point-source emissions not included in bottom-up inventories by redistributing oil-gas and coal emissions based on a gridded inventory constructed from GHGSat plume and null detections, and by enforcing native resolution emissions optimization at locations where plumes were observed by point source imagers including PRISMA, Sentinel-2, Landsat, EnMAP, GOES, and EMIT, and where large plumes were detected by TROPOMI. Our global total posterior emission of 562 Tg for 2023 is in line with previous coarse-scale global inversion studies. The inversions’ high resolution allows source separation and independent optimization of individual countries, confirmed by small posterior error correlations between countries. China (52.7 Tg), the U.S. (32.2 Tg), India (25.7 Tg), Brazil (18.5 Tg), and Indonesia (10.7 Tg) have the highest anthropogenic emissions, representing 14%, 9%, 7%, 5%, and 3% of the global total anthropogenic source, respectively. Uncertainty estimates come from an inversion ensemble with varied inversion parameters. Results provide an estimate of emissions from all countries in a globally consistent inverse modeling framework, serve as a direct comparison and aid of countries’ UNFCCC reporting, and provide up-to-date observational constraints on emissions from countries where reporting is unfeasible or out of date.