This report defines the necessary components for a multi-scale and multi-species data assimilation (DA) system that targets anthropogenic CO₂ emissions. This DA system will use multiple streams of observations, including satellite observations. Out of the many viable options to serve as the basis for such a system on the global scale, we see a hybrid 4d-VAR-ensemble approach, implemented in an online transport model, and operated within a Numerical Weather Prediction environment, as a fundamental building block. On top of this a DA system should use multiple tracers, be adoptable to long- and short windows, and optimize both the atmospheric state as well as surface fluxes.

Such a system does currently not yet exist, and we recommend a number of concrete research and development needs, including to:

• Allow mass-conserving transport in the operational Integrated Forecast System (IFS) of ECMWF
• Improve the treatment of background covariances and building of long-window information transfer in the IFS DA system
• Develop Fossil Fuel DA models and capacity (FFDAS) for global and regional scales
• Expand Biospheric Carbon Cycle DA models and capacity (CCDAS) for global and regional scales
• Invest in multi-tracer transport+source modelling on all scales
• Improve the seamless coupling of regional DA systems to the global IFS
• Investigate plume-based methods for fast DA, also in a plume-in-grid approach

Many of these developments are ongoing in the community and to facilitate their uptake in an MVS for anthropogenic CO₂, we see an important role for:

1. a prototype MVS built around the IFS and focusing on available high-resolution CO and NO₂ satellite data
2. a multi-scale integration tool that allows local- and regional scale DA systems to feed into the global analyses.