This report presents the high dimensional inversion framework designed for the co-assimilation of CO2 (from satellite or ground-based measurements) and additional tracers of fossil fuel emissions (CO, 14CO2 ground-based observations) and separately controlling emissions from large industrial plants, cities and regional budgets of more diffuse emissions. It currently represents most of north-eastern France, Benelux and western Germany with a ~ kilometric resolution. Results are illustrated here about the assimilation of satellite retrievals alone, about the assimilation of ground-based CO2 and 14CO2 measurements, and about the sizeable impact of model error. A larger suite of results is presented and discussed in D4.4, “Sampling Strategy for additional tracers”, in order to contribute to the design of the surface network.
The high-resolution control vector avoids over-optimistic assessment of the capability of observation systems, but involves a very large computational burden. The assimilation of CO and 14CO2 is considered separately in order to optimize the computational effort for each tracer. For the CO-CO2 inversion configuration, we prioritize the control of the different sectors of anthropogenic activity emitting both CO2 and CO for each target area but we focus on Belgium only. For the 14CO2-CO2 inversion configuration, we do not need the sectoral resolution and can put the priority on the distinction between various regions around Belgium in the control vector.
Technically, the system requires high performance computing and has been run on a French supercomputer. A 14CO2-CO2 inversion configuration without radiocarbon involves about ten thousand 24-hour response functions for a 24-hour inversion window. The computation of each response function requires a couple of “wall clock” hours, while other computations in the systems take less than 5 “wall clock” hours. The overall performance could be further optimized for operational use within the future anthropogenic CO2 emissions Monitoring and Verification Support capacity (CO2MVS).