Methane (CH4) emissions have to be reduced to reach the Paris Agreement objectives. Modeling approaches assimilating observations are expected to verify the effectiveness of mitigation policies by monitoring sectoral and regional emission changes based on atmospheric observations. These “top-down” approaches are appropriate to estimate CH4 net flux at the surface; however, they have critical weak points: an incomplete prior knowledge of CH4 natural emissions, and difficulties in disentangling overlapping sources from different sectors. The main natural sources of CH4 are inland water emissions: wetlands, peatlands, lakes, ponds… While global wetland emissions gridded products exist, none are available for other inland water systems, which are thus missing in the prior description of emissions forcing top-down approaches. Global estimates of inland water systems remain not consistent in terms of localization and magnitude. Thus, dynamical maps of these areas and their emissions based on same data sets, are necessary to ensure consistency and reduce uncertainties in CH4 emission estimates. 3D top-down approaches retrieving CH4 emissions are based on total CH4 observations only, at surface stations or from remote sensing, and solve for net fluxes at the surface or few individual sectors. However, CH4 observations are not sufficient to properly distinguished between sectors. Isotopic or co-emitted species such as ethane for oil and gas emissions, or carbon monoxide for biomass burning, could help better discriminating the emissions from different sectors and their trend. For this, the project coordinator’s team has newly developed an inverse system (CIF-LMDz-SACs) that integrates isotopic observations to constrain CH4 emissions. The AMB-M3 project aims at 1) producing the first consistent gridded product of CH4 emissions from wetlands, peatlands and lakes to be used for CH4 sources analysis and as input to global top-down models, and 2) discriminating CH4 sources by improving the capabilities of an inverse systems. In the first work package we will develop dynamical and consistent gridded CH4-centric inland water maps including information on vegetation and soil types (carbon content), over 1995-2020 based on the GIEMS-2 inundated data set. GIEMS-2, covering already 1995-2015, provides inundated extent and dynamics under dense vegetation, produces consistent surface water in semi-arid regions and accounts for satellite inter-calibration issues. Then CH4 emissions based on this new product will be estimated by upscaling density fluxes from the literature. Developments on CH4 emissions from wetlands and northern peatlands have been done in different versions of the ORCHIDEE land surface model. For the project, they will be integrated into the trunk version of the ORCHIDEE model to better simulate CH4 emissions and compared to previous estimations. The second work package will demonstrate the potential of adding ethane and carbon monoxide within the CIF-LMDz-SACs surface data constrained global inversions. Analyses will be performed to optimally combine different tracer constraints from both surface and satellite observations to take advantage of all available observations for regional analysis. Finally, AMB-M3 project will provide new estimates of CH4 sources and sinks at the global and regional scales over 2010-2020 based on up-to-date bottom-up (ORCHIDEE) estimates and top-down modelling system.